energies Article

A Photovoltaic-Based SEPIC Converter with Dual-Fuzzy Maximum Power Point Tracking for Optimal Buck and Boost Operations Tanaselan Ramalu 1,2, *, Mohd Amran Mohd Radzi 1,2 , Muhammad Ammirrul Atiqi Mohd Zainuri 1,2 , Noor Izzri Abdul Wahab 1,2 and Ribhan Zafira Abdul Rahman 1 1

2

*

Department of Electrical and Electronic Engineering, Faculty of Engineering, University Putra Malaysia, 43400 Serdang, Selangor, Malaysia; [email protected] (M.A.M.R.); [email protected] (M.A.A.M.Z.); [email protected] (N.I.A.W.); [email protected] (R.Z.A.R.) Centre for Advanced Power and Energy Research, Faculty of Engineering, University Putra Malaysia, 43400 Serdang, Selangor, Malaysia Correspondence: [email protected]; Tel.: +60-17-2566-574

Academic Editor: Gabriele Grandi Received: 17 May 2016; Accepted: 27 July 2016; Published: 30 July 2016

Abstract: In this paper, a photovoltaic (PV)-based single ended primary-inductor converter (SEPIC) is developed with introduction of dual-fuzzy logic controller (FLC) maximum power point tracking (MPPT) algorithm. Separate FLC parts, for the first time used for MPPT, are configured for optimal operations of both buck and boost operations. During buck operation, a high overshoot voltage exists, and during boost operation, an undershoot voltage occurs, both during the initial rising period. Definitely, a single-FLC MPPT could not be able to minimize both problems, which on the other hand can be handled by the proposed MPPT algorithm. For evaluation purposes, buck operation has been conducted during high irradiance, while during low irradiance, boost operation has been conducted. The dual-FLC MPPT with SEPIC was simulated in MATLAB-Simulink, and further a laboratory prototype was implemented with a TMS320F28335 eZdsp board. Both simulation and experimental results and comparison analysis (with the single-FLC MPPT) have been presented. From the results and analysis, the dual-FLC MPPT performs better than the single-FLC MPPT in terms of faster response time, lower overshoot and undershoot, and further significant reduction of power losses. Keywords: photovoltaic (PV); maximum power point tracking (MPPT); fuzzy logic controller (FLC); single ended primary-inductor converter (SEPIC)

1. Introduction In terms of renewable energy, there is a long list of energy types that comes from various natural resources such as solar, wind, geothermal, sea tide, and biomass. Among them, photovoltaic (PV) from solar is much preferable due to its implementation simplicity with less maintenance. In recent years, PV systems have witnessed neverending demand due to their enormous potential to be the nearest solution we have right now to substitute our diminishing fossil fuel energy sources. When a PV panel is exposed to solar irradiation, it can generate direct current electricity without any environmental impact or contamination. The only drawbacks are that the cost to manufacture PV panels is too high and their small range of efficiency is only about 15%–20% [1–3]. PV panels have nonlinear output characteristics and the main factors affecting PV output power are the solar irradiation, temperature and load impedance [3]. When the solar irradiation rises, the PV current increases, however, the temperature of a PV module has a more significant effect on PV voltage operation [4]. Due to the nonlinear output characteristics of PV panels, an algorithm is needed to track Energies 2016, 9, 604; doi:10.3390/en9080604

www.mdpi.com/journal/energies

Energies 2016, 9, 604

2 of 17

the maximum power point (MPP) of the PV curve to deliver the maximum power. This is known as maximum power point tracking (MPPT). The MPPT operation basically involves finding the maximum operating voltage and current at which PV operates to achieve the MPP. Many MPPT methods have been developed and implemented [4,5]. Among them are perturb and observe (P&O), incremental conductance (IC), artificial neural network (ANN), fuzzy logic controller (FLC), constant voltage, three-point weight comparison, short current pulse, and open circuit voltage. The most commonly used traditional methods are P&O and incremental conductance, however, potential artificial intelligent techniques like FLC and ANN are recently gaining popularity in MPPT design due to their ability to achieve higher stability and less noise factor [6]. Specifically, FLC does not require an accurate mathematical model and is known to be very efficient in handling problems that have non-linear variables [6]. Meanwhile, the amount of DC-based equipment operated at various levels is growing higher, and thus, having a DC-DC converter with ability to produce various DC outputs is preferable. The SEPIC converter is preferable due to its ability to buck and boost input voltage and has advantage of having non-inverted output. By having non-inverted output polarity, implementation of circuit to load becomes easier due to the fact the reference point (ground) is the same. There are quiet a number of significant works on SEPIC with PV [7], which also covers MPPT algorithms, including FLC [8]. As MPPT tracks maximum power via increasing or decreasing voltage and current, SEPIC is effective as it increases and decreases voltage at the current’s expense [9]. However, in related to FLC MPPT for SEPIC, there are no such works considering comprehensive evaluations of MPPT for both buck and boost operations. By using only a single FLC for MPPT, although it performs better as compared to P&O [10], using just a common pattern of membership functions may degrade its performance to track MPP, especially when facing dynamic irradiance changes. In addition, at the controller output for producing a PWM signal to power switching devices in SEPIC, this type of MPPT is only suitable to be set as a duty cycle change which will be added to a pre-defined duty cycle. Use of a direct duty cycle approach is totally unsuitable. Consequently, when the irradiance changes rapidly, there is a possibility that the single-FLC MPPT will fail to track its MPP and take a certain amount of time to reach steady state conditions [11,12]. Therefore, this paper proposes a dual-FLC MPPT that offers a significant improvement of both buck and boost operations in SEPIC. Design of membership functions for each FLC will be carried out by focusing to each specific problem in buck and boost operations, respectively. Hypothetically, as further proven later in this paper, during buck operation, a high overshoot voltage exists, and during boost operation, an undershoot voltage occurs, both during the initial period of changes. Thus, the proposed MPPT should ensure not only operation of SEPIC at MPP, but also should address the mentioned problems, both in simulation and experimental works. Two separate loads with a switching circuit, called as load changing circuit, are set up to ensure a significant impact of dynamic changes can be provided during evaluation of the proposed MPPT. As for the rest of this paper, Section 2 covers the proposed PV-based system, followed by a description of the proposed MPPT in Section 3, and both simulation and experimental validations in Sections 4 and 5, respectively. Finally Section 6 concludes the findings. 2. Proposed Photovoltaic-based SEPIC System The proposed PV-based SEPIC system, as shown in Figure 1, consists of five main parts: PV panel, SEPIC, controller which consists of dual-FLC MPPT and load changing algorithms, load changing circuit, and loads. Meanwhile, Figure 2 presents the configuration of SEPIC. Voltage and current are measured and used by the dual-FLC MPPT to produce a suitable duty cycle to operate IGBT in SEPIC.

Energies 2016, 9, 604 Energies 2016, 9, 604

3 of 17 3 of 17

Load changing circuit

Sepic converter

PV Panel

Load 1

Duty cycle V

I DUAL FLC MPPT

Load 2

Load changing algorithm DSP

Figure proposedPV-based PV-basedSEPIC SEPICsystem. system. Figure1.1.Block Block diagram diagram of of proposed Inductor 1

Capacitor 1

Diode

PV panel Input Capacitor

IGBT

Inductor 2 Capacitor 2

Load

Figure2.2.Configuration Configuration of Figure of SEPIC SEPIC with withPV PVasasinput. input.

The voltage conversion ratio of SEPIC can be defined as follows: The voltage conversion ratio of SEPIC can be defined as follows:

𝐷 𝑉𝑜𝑢𝑡 =ˆ( D ˙) 𝑉𝑖𝑛 Vout “ 1 − 𝐷 Vin

(1)

(1) 1´D where Vout and Vin are the output and input voltages of SEPIC, respectively, and D is the duty cycle which is defined thethe ratio of theand turn-on the switching time period. where Vout and Vinas are output inputduration voltagestoof SEPIC, respectively, and D is the duty cycle Two separate loads are connected through a load changing circuit, considering one load is which is defined as the ratio of the turn-on duration to the switching time period. specifically used for buck operation and another one for boost operation. Critical performance during Two separate loads are connected through a load changing circuit, considering one load is the switching between buck boost one operations be investigated A threshold specifically used period for buck operation andand another for boostcan operation. Critical later. performance during current is set as 3 A for the load changing algorithm. If the current is less than the threshold value, the switching period between buck and boost operations can be investigated later. A threshold current the as circuit willthe connect SEPIC to load 1. OnceIfthe current is more thanthreshold the threshold value, the is set 3 A for load changing algorithm. theinput current is less than the value, the circuit circuit will connect SEPIC to load 2. When irradiance is low, the current delivered from the PV panel will connect SEPIC to load 1. Once the input current is more than the threshold value, the circuit will will be lower and SEPIC performs boost operation with FLC 1. When the irradiance is higher, the connect SEPIC to load 2. When irradiance is low, the current delivered from the PV panel will be lower input current from the PV panel will increase, so SEPIC will perform buck operation with FLC 2. and SEPIC performs boost operation with FLC 1. When the irradiance is higher, the input current from the3.PV panel willLogic increase, so SEPIC will perform Dual-Fuzzy Controller Maximum Powerbuck Pointoperation Tracking with FLC 2. As mentioned, FLCs areMaximum used as MPPT in this system to perform buck and boost operations, 3. Dual-Fuzzy Logictwo Controller Power Point Tracking respectively. The same two inputs (error E and change of error CE) at sample time k are used, which As mentioned, two FLCs are used as MPPT in this system to perform buck and boost operations, are defined as below: respectively. The same two inputs (error E and change of error CE) at sample time k are used, which 𝑃(𝑘)−𝑃(𝑘−1) are defined as below: (2) E (k) = P pkq ´ P pk ´ 1q E pkq “ 𝑉(𝑘)−𝑉(𝑘−1) (2) V pkq ´ V pk ´ 1q CE (k) = E (k) − E (k − 1) (3) CE pkq “ E pkq ´ E pk ´ 1q (3) where E is the change of PV power over the change of PV voltage, and CE is the difference between where E is theEchange PV power the sample change time. of PV voltage, and CE is the difference between the current from theofprevious E atover a given Basically, thethe operation of E FLC be classified into four main elements: fuzzification, rule base, the current E from previous at acan given sample time. inference engine defuzzification [13–17]. During fuzzification, the inputs of thefuzzification, FLC, CE and rule E Basically, the and operation of FLC can be classified into four main elements: variables are calculated anddefuzzification converted into linguistic based on the membership base, inference engine and [13–17]. variables During fuzzification, the inputs functions. of the FLC, The output (in this case it is the duty cycle D) is generated by looking up in a rule base table [18]. The CE and E variables are calculated and converted into linguistic variables based on the membership fuzzy output is converted back to a numerical variable from a linguistic variable during functions. The output (in this case it is the duty cycle D) is generated by looking up in a rule base

Energies 2016, 9, 604 Energies 2016, 9, 604

4 of 17 4 of 17

Energies 2016, 9, 604

4 of 17

table [18]. The fuzzy output converted backFLCs to a numerical from a linguistic variable defuzzification [19–23]. Theisdesign of both differs invariable their patterns and ranges, as to during reflect defuzzification [19–23]. The design of both FLCs differs in their patterns and ranges, as to reflect effectiveness of buck and boost operations, respectively. The concept of designing them is based defuzzification [19–23]. The design of both FLCs differs in their patterns and ranges, as to reflect on effectiveness buck and boost operations, respectively. The designing them based the PV curve,ofand mapping ofoperations, membership functions inconcept the PVof curve for the proposed MPPT is effectiveness of the buck and boost respectively. The concept of designing them isisbased onon the thein PVFigure curve, and the mapping of membership functions thecurve PV curve for the proposed MPPT shown PV curve, and the3. mapping of membership functions in theinPV for the proposed MPPT is is shown shown in Figure 3. in Figure 3. Power(P) Power(P)

Voltage(V)

Voltage(V) Figure 3. Mapping of the membership functions in the P-V curve for dual-FLC MPPT.

Figure 3. Mapping of the membership functions in the P-V curve for dual-FLC MPPT. Figure 3. Mapping of the membership functions in the P-V curve for dual-FLC MPPT. The PV curve has been divided into seven segments with each representing different The hashas beenin divided into seven with eachwith representing different membership membership functions the dual-FLC MPPT. The seven membership functions are Negative Big The PV PVcurve curve been divided into segments seven segments each representing different (NB), Negative Medium (NM), Negative Small (NS), Zero (ZE), Positive Small (PS), Positive Medium functions in the dual-FLC MPPT. The seven membership functions are Negative Big (NB), Negative membership functions in the dual-FLC MPPT. The seven membership functions are Negative Big and Positive Big (NM), (PB). ZE membership function is located exactly at the maximum power point Medium (NM), Negative Small (NS), Zero (ZE), Positive Small (PS), Positive Medium (PM) and (NB),(PM) Negative Medium Negative Small (NS), Zero (ZE), Positive Small (PS), Positive Medium of the PV curve. The membership functions of the left hand side of ZE (PS, PM, PB), will be labelled Positive (PB). ZE function function is located the maximum power point of the (PM) andBig Positive Big membership (PB). ZE membership is exactly located at exactly at the maximum power point as positive polarity as the gradient of the PV curve is positive, and the membership functions of the PV curve. The membership functions of the left hand side of ZE (PS, PM, PB), will be labelled as of the PV curve. The membership functions of the left hand side of ZE (PS, PM, PB), will be labelled right hand side of ZE (NS, NM, NB) will be labelled as negative polarity as the gradient of the PV positive polarity as the gradient of the is positive, and the functions of theofright as positive polarity as the gradient of PV the curve PV curve is positive, andmembership the membership functions the curve is negative. The areas of PS and NS for both polarities, in which the controller becomes more hand side of ZE (NS, NM, NB) will be labelled as negative polarity as the gradient of the PV curve is rightsensitive hand side of ZEZE(NS, NM, NB) willduty be labelled as controller, negative are polarity as the gradient of the PV towards which determines cycle of the critical. negative. The areasThe of PS both in which thewhich controller becomes sensitive curve is negative. areas ofNS PSfor and NS polarities, for both polarities, in controller becomes As mentioned, theand design of membership functions for both FLCs the is different, as more shown in more towards ZE which determines duty cycle of the controller, are critical. sensitive towards ZE which determines duty cycle of the controller, are critical. Figures 4 and 5, respectively. FLC 1 is specially designed to overcome undershoot voltage problems As the design of membership functions voltage for both FLCs is In different, as seven shown in and mentioned, FLC 2 is specially designed to overcome overshoot problems. each FLC, membership functions areFLC configured for all inputs and output. All theundershoot membership functions are set and Figures 44and 1 is specially designed to overcome voltage problems Figures and5,5,respectively. respectively. FLC 1 is specially designed to overcome undershoot voltage problems shapes with both ending sides of the universe disclosure accompanied a trapezium FLC istriangular specially designed to overcome overshoot voltageofproblems. each FLC,Inby seven membership and 2as FLC 2 is specially designed to overcome overshoot voltage In problems. each FLC, seven shapeare to show continuous operation controller. The selected fuzzy is Mamdani functions configured forconfigured all inputs of and output. the membership functions are functions set aswhere triangular membership functions are forthe all inputsAll and output. All the method membership are set the maximum of the minimum composition technique for the inference is used. The center-of-gravity shapes with both ending the universe disclosure a trapeziumby shape to show as triangular shapes withsides bothof ending sides ofofthe universeaccompanied of disclosureby accompanied a trapezium method is used in the defuzzification process.

Degree of membership

Degree of membership

continuous operation of theoperation controller.ofThe fuzzy is fuzzy Mamdani where the maximum of shape to show continuous theselected controller. Themethod selected method is Mamdani where the technique for the inference used. center-of-gravity is used in the minimum maximumcomposition of the minimum composition techniqueisfor the The inference is used. Themethod center-of-gravity NB NM NS ZE PS PM PB the defuzzification process. method is used1 in the defuzzification process. NB

1 0.5

0.5

NM

NS ZE

PS

PM

PB

0

-2.0

-1.5

-1

-0.5

0

0.5

1

1.5

2

Error, E (ΔP/ΔV)

(a) 0

-2.0

-1.5

-1

-0.5

0

Error, E (ΔP/ΔV)

(a) Figure 4. Cont.

0.5

1

1.5

2

membership Degree membership ofof Degree

Energies 2016, 9, 604 Energies 2016, 9, 604 Energies 2016, 9, 604

5 of 17 5 of 17 5 of 17 NB NB

1 1

NM NM

NS ZE NS ZE

PS PS

PM PM

PB PB

0.5 0.5

0 0 -2.0 -2.0

-1.5 -1.5

-1 -1

-0.5 -0.5

0 0

0.5 0.5

Change of Error, CE (ΔE) Change of Error, CE (ΔE)

1 1

1.5 1.5

2 2

membership Degreeofofmembership Degree

(b) (b)

NB NB

1 1

NM NM

NS NS

ZE ZE

PS PS

PM PM

PB PB

0.5 0.5

0 00.2

0.3 0.3

0.2

0.4 0.4

0.5 0.5

0.6 0.6

Output, ΔD Output, ΔD

0.7 0.7

0.8 0.8

0.9 0.9

1 1

(c) (c)

membership Degree membership ofof Degree

Figure 4. Fuzzy membership functions of inputs and output for FLC 1: (a) error; (b) change of error; Figure functions of of inputs inputs and and output output for for FLC FLC 1: 1: (a) (a) error; error; (b) (b) change change of of error; error; Figure 4. 4. Fuzzy Fuzzy membership membership functions and (c) output. and (c) output. and (c) output. 1 1

NB NB

NM NM

NS ZE NS ZE

PS PM PS PM

PB PB

0.5 0.5

0 0-2.0

-1.5 -1.5

-2.0

-1 -1

-0.5 -0.5

0 0

Error, E (ΔP/ΔV) Error, E (ΔP/ΔV)

0.5 0.5

1 1

1.5 1.5

2 2

1.5 1.5

2 2

membership Degree membership ofof Degree

(a) (a)

1 1

NB NB

NM NM

NS ZE NS ZE

PS PM PS PM

PB PB

0.5 0.5

0 0-2.0

-2.0

-1.5 -1.5

-1 -1

-0.5 -0.5

0 0

0.5 0.5

Change of Error, CE (ΔE) Change of Error, CE (ΔE) (b) (b) Figure 5. Cont.

1 1

Energies 2016, 9, 604 Energies 2016, 9, 604

6 of 17 6 of 17

Degree of membership

NB

NM

NS

ZE

PS

PM

PB

1

0.5

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Output, ΔD

(c) Figure Figure 5. 5. Fuzzy Fuzzy membership membership functions functions of of inputs inputs and and output output for for FLC FLC 2: 2: (a) (a) error; error; (b) (b) change change of of error; error; and (c) output of FLC 2. and (c) output of FLC 2.

In FLC 1, the negative polarity membership functions such as NB, NM and NS are arranged In FLC 1, the negative polarity membership functions such as NB, NM and NS are arranged closely to ZE so that the right hand side of the PV curve is given more priority because the selected closely to ZE so that the right hand side of the PV curve is given more priority because the selected area covered under the curve has higher voltage and this could compensate the undershoot voltage. area covered under the curve has higher voltage and this could compensate the undershoot voltage. Meanwhile, in FLC 2, the positive polarity membership functions such as PB, PM and PS are arranged Meanwhile, in FLC 2, the positive polarity membership functions such as PB, PM and PS are arranged closely to ZE so that the left hand side of the PV curve is given more priority because the selected closely to ZE so that the left hand side of the PV curve is given more priority because the selected area area covered under the curve has lesser voltage and this could reduce the overshoot voltage. covered under the curve has lesser voltage and this could reduce the overshoot voltage. Switching Switching signals generated by SEPIC are based on direct duty, so as to overcome the limitation of signals generated by SEPIC are based on direct duty, so as to overcome the limitation of the single the single FLC mentioned early. However, an initial D has separately been set for FLC 1 and FLC 2 FLC mentioned early. However, an initial D has separately been set for FLC 1 and FLC 2 patterns patterns respectively for preventing the output voltage to have unwanted overshoot or undershoot respectively for preventing the output voltage to have unwanted overshoot or undershoot which could which could harm the system. Table 1 shows the 49 rules for both FLCs. harm the system. Table 1 shows the 49 rules for both FLCs. Table 1. Rules used for both FLCs. Table 1. Rules used for both FLCs.

E/CE E/CE NB NM NB NM NS NS ZE ZE PS PS PM PM PB PB

NB NB ZE ZE ZE ZE ZE ZE PS PS NS NS NM NM NB NB

NM NS ZE NM NS ZE ZE ZE PB ZE ZE PM ZE ZE PB ZE ZE ZE PM PS ZE ZE PS ZE ZE ZE ZE ZE ZE NS NS NS NS NS NS NM NM NM NM NM NM NB NB NB NB NB NB

PS

PM

PB

PS PB

PM PB

PB PB

PM PB PM PB PM PB PM PS PM PS PM PS PS PS PS ZE ZE NS ZE ZE NS ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE

Finally, full operation operation of of the the controller controller is is shown shown in in Figure Figure 6. 6. After Finally, full After measuring measuring the the PV PV voltage voltage and and current, current, power power will will be be calculated. calculated. Then, Then, changes changes of of PV PV power power and and voltage voltage are are calculated. calculated. After After that, that, error error E E and and change change of of error error CE CE are are determined determined for for later later use use as as inputs inputs for forthe theselected selectedFLC. FLC.Meanwhile, Meanwhile, FLC and load must be chosen, and this process considers measurement of input current. As FLC and load must be chosen, and this process considers measurement of input current. As mentioned, mentioned, when the input current is lower than the threshold value (which is 3 A) due to low when the input current is lower than the threshold value (which is 3 A) due to low irradiance level, irradiance and load 1 will be selected. when the input current is highervalue, than FLC 1 and level, load 1FLC will1be selected. However, when However, the input current is higher than threshold threshold load 2 will be selected. FLC 2 andvalue, load 2FLC will 2beand selected.

Energies 2016, 9, 604

7 of 17

Energies 2016, 9, 604

7 of 17 Start

Measure V(k) and I(k)

Calcu late P(k)

ΔP = P(k) – P(k-1) ΔV = V(k) – V(k-1)

E(k) = ΔP/ΔV CE(k) = E(k) – E(k-1)

NO

FLC 1 and Load 1 (Boos t)

YES

Input current > 3 A

FLC 2 and Load 2 (Buck)

Figure 6. Flow chart of the controller. Figure 6. Flow chart of the controller.

4. Simulation Results 4. Simulation Results The PV-basedSEPIC SEPICsystem systemwas wasfirst first modeled MATLAB/Simulink. In this work, The proposed PV-based modeled in in MATLAB/Simulink. In this work, the 2 (KD210GH2 the selected PV panel is a polycrystalline silicon type that produces 210 W at 1000 W/m selected PV panel is a polycrystalline silicon type that produces 210 W at 1000 W/m (KD210GH-2PU, 2PU, Kyocera, Esslingen, Germany). Theparameters main parameters of thePV selected PV panel have been Kyocera, Esslingen, Germany). The main of the selected panel have been tabulated in tabulated in Table 2. Meanwhile, the main components and parameters of shown SEPIC are shown Table 2. Meanwhile, the main components and parameters of SEPIC are in Table 3. in Table 3. For boost operation, the initial duty cycle is set to 0.6 with output voltage will be 39.9 V, and the Tabledown 2. Parameters Kyocera KD210GH-2PU. output voltage will be stepped to 17.7 of V the in buck operation with duty cycle of 0.4. The lower and higher irradiances are set to 200 W/m2 and 700 W/m2., and from both irradiances, the expected Item Value currents produced by the PV system are 1.2 A and 5.63 A. Maximum power (Pmpp ) 210 W Maximum operating (VmppKD210GH-2PU. ) 26.6 V Table 2. Parameters ofvoltage the Kyocera Maximum operating current (Impp ) 7.9 A Short circuit current (Isc ) 8.58Value A Item Open circuit voltage (V ) 33.2 V W oc Maximum power (Pmpp) 210 Temperature coefficient of open circuit voltage ´0.36%/K Maximum operating voltage (Vmpp) 26.6 V Temperature coefficient of short circuit current 0.06%/K Maximum operating current (I mpp ) 7.9 A Temperature coefficient of maximum power ´0.46%/K ˝ Normal operating 25 8.58 C A Short circuit currentcell (Isc)temperature

Open circuit voltage (Voc) Tableof3.open Maincircuit components of SEPIC. Temperature coefficient voltage Temperature coefficient of short circuit current Item Temperature coefficient of maximum powerValue Normal operating cell temperature Inductors 1 and 2 3 mH

33.2 V −0.36%/K 0.06%/K −0.46%/K 25 °C

Capacitors 1 and 2 1000 µF Input 470 µF Table 3. capacitor Main components of SEPIC. Switching frequency 25 kHz Loads 1 and 2 50 Ω and 2 Ω Item Value

Inductors 1 and 2 3 mH Capacitors 1 and 2 to 0.6 1000 F For boost operation, the initial duty cycle is set withµoutput voltage will be 39.9 V, and the Input 470 µ F with duty cycle of 0.4. The lower output voltage will be stepped down to capacitor 17.7 V in buck operation Switching frequency 25 kHz Loads 1 and 2 50 Ω and 2 Ω

Energies 2016, 9, 604

8 of 17

and higher irradiances are set to 200 W/m2 and 700 W/m2 ., and from both irradiances, the expected Energies 2016, 9, 604 8 of 17 currents produced by the PV system are 1.2 A and 5.63 A. For purpose of comparative evaluation, a single-FLC MPPT has been developed too [21]. For purpose of comparative evaluation, a single-FLC MPPT has been developed too [21]. The The membership functions of the single FLC are shown in Figure 7. The membership functions membership functions of the single FLC are shown in Figure 7. The membership functions must must equally be arranged, as to address bothand buck andfunctions. boost functions. As mentioned, output is equally be arranged, as to address both buck boost As mentioned, its output its is changed changed duty to belater added later the pre-defined duty cycle. by dutyby cycle tocycle be added with thewith pre-defined duty cycle. NB

NM

NS

ZE

PS

PM

-2

-1

0

1

2

PB

Degree of membership

1

0.5

0

-5

-4

-3

3

4

5

Error, E (ΔP/ΔV)

(a)

Degree of membership

NB

NM

NS

ZE

PS

PM

-2

-1

0

1

2

PB

1

0.5

0

-5

-4

-3

3

4

5

Change of Error, CE (ΔE)

(b) NB

NM

NS

-2

-1

ZE

PS

PM

0

1

2

PB

Degree of membership

1

0.5

0

-5

-4

-3

Output, ΔD

3

4

5

(c) Figure 7. Membership functions for single-FLC MPPT: (a) error; (b) change of error; and (c) output. Figure 7. Membership functions for single-FLC MPPT: (a) error; (b) change of error; and (c) output.

To further consolidate the contribution of this paper, Dual FLC has been compared with another further consolidate ofof this paper, Dual FLC hasSingle been compared another twoTomore single FLCs withthe thecontribution same number membership functions, FLC 1 and with Single FLC two more single FLCs with the same number of membership functions, Single FLC 1 and Single FLC 2 2 MPPT that have different membership function patterns. Single FLC 1 has membership functions MPPT that have different membership function patterns. Single FLC 1 has membership functions patterned similar to FLC 1 as in Figure 4, while Single FLC 2 has membership functions similar to patterned similar to FLC 1 as in Figure while FLC Single FLC Single 2 has membership functions similar8 to FLC FLC 2 as in Figure 5. The design of4,Single 1 and FLC 2 is shown in Figures and 92 as respectively. in Figure 5. The design of Single FLC 1 and Single FLC 2 is shown in Figures 8 and 9 respectively.

Energies2016, 2016,9,9,604 604 Energies

99ofof17 17

Degree of membership

Energies 2016, 9, 604

9 of 17

NB

Degree of membership

1

NM

NS ZE

PS

NM

NS ZE

PS

NB

1

PM

PB

PM

PB

0.5 0.5

0

-2.00

-1.5

-2.0

-1

-1.5

-1

-0.5

0

-0.5

Error, E0(ΔP/ΔV) Error, E (ΔP/ΔV) (a)

0.5

1

0.5

1.5

1

2

1.5

2

NB

1

Degree of membership

Degree of membership

(a)

NM

NB

1

NM

NS ZE

PS

NS ZE

PM

PS

PM

PB

PB

0.50.5

0 0

-2.0-2.0

-1.5 -1.5

-1-1

-0.5 -0.5

00

0.5 0.5

11

1.51.5

2 2

Change of of Error, Error,CE CE(ΔE) (ΔE)

Degree of membership

Degree of membership

(b) (b) NB NB

1 1

NM NM

NS NS

ZE ZE

PS PS

PM PM

PBPB

0.5

0.5

0

0

0.2

0.3

0.2

0.3

0.4

0.5

0.4

0.5

0.6

0.7

0.6

0.8

0.7

Output, ΔD

0.9

0.8

1

0.9

1

Output, ΔD (c)

(c)

Figure 8. Fuzzy membership functions of inputs and output for single FLC 1: (a) error; (b) change of

Degree of membership

Degree of membership

Figure 8. Fuzzy membership functions of inputs and output for single FLC 1: (a) error; (b) change of Figure Fuzzy membership error;8.and (c) output of singlefunctions FLC 1. of inputs and output for single FLC 1: (a) error; (b) change of error; error;and and(c) (c)output outputofofsingle singleFLC FLC1.1. 1

1

NB

NM

NB

NM

NS

ZE

PS

NS

ZE

PS

PM

PB

PM

PB

0.5

0.5

0

-2.0

-1.5

-1

-2.0

-1.5

-1

0

-0.5

0

-0.5

Error, E (ΔP/ΔV) 0

Error,(a)E (ΔP/ΔV)

(a) Figure 9. Cont.

0.5

0.5

1

1.5

1

1.5

2

2

Energies 2016, 9, 604 Energies Energies2016, 2016,9,9,604 604 NB NB

11

membership of membership Degree of Degree

10 of 17 1010ofof1717 NM NM

NS NS

ZE ZE

PS PM PM PS

PB PB

0.5 0.5

00

-2.0 -2.0

-1.5 -1.5

-1 -1

0 0

-0.5

0.5 0.5

11

1.5 1.5

22

Change of Error, Error, CE CE (ΔE) (ΔE)

(b)

Degree membership ofmembership Degreeof

NB NB

NM NM

NS

ZE ZE

PS PS

PM PM

PB PB

11

0.5 0.5

00

00

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4

Output, Output, ΔD ΔD

0.5 0.5

0.6 0.6

0.7 0.7

0.8 0.8

(c) (c) Figure 9. Fuzzy membership functions of inputs and output for single FLC 2: (a) error; (b) change of Figure 9. Fuzzy membership functions of inputs and output for single FLC 2: (a) error; (b) change of error; and (c) output of single FLC 2. output of of single single FLC FLC 2. 2. error; and (c) output

Input Power(W) Input Power(W)

The simulation work initially considers measurement and analysis at PV input due to effect of The simulation simulation work work initially considers considers measurement and at PV input due to effect effect of The measurement and analysis analysis PV both inputloads due to changed irradiance levels.initially In addition, performance of output voltagesatfor has beenof changed irradiance irradiancelevels. levels. In addition, performance of output voltages for both has been changed In addition, performance of output voltages for both hasloads been analyzed analyzed by undergoing switching operation by load changing circuit. The loads waveforms have been analyzed by undergoing switching operation by load changing circuit. The waveforms have been by undergoing operation load changing The waveforms been captured, captured, andswitching performance of thebydual-FLC MPPTcircuit. is compared with thehave single-FLC MPPT, and by captured, and performance of theisdual-FLC MPPT is single-FLC compared MPPT, with the single-FLC parameters MPPT, by performance of the dual-FLC MPPT compared with the by considering considering parameters such as response time and power loss, both during transient part. Figure 10 considering parameters such as response time andtransient power loss, both during transient part. Figure 10 such as response time power loss, Figure 10 shows waveforms of both input shows waveforms of and input powers forboth bothduring buck and boost part. operations. During buck operation, shows waveforms of input powers for both buck and boost operations. During buck operation, both powers both buck and boost During buck operation, bothatMPPTs toduring control MPPTsfor manage to control SEPICoperations. to achieve maximum power as expected 155 W. manage However, MPPTs manage to control SEPIC to achieve maximum power as expected at 155 W. However, during SEPIC to achieve maximum expectedmuch at 155better W. However, duringtime transient the transient part, the proposedpower MPPTasperforms with response only part, 0.1 ms toproposed achieve transient part, the proposed MPPT performs much better with response time only 0.1 ms to achieve MPPT bettertowith time only 0.1 ms0.4 to achieve steady state lower as compared the steadyperforms state as much compared the response single-FLC MPPT with ms. Consequently, energytoloss steady stateMPPT as compared to the single-FLC lower MPPTenergy with 0.4 ms. Consequently, lower energy loss single-FLC with 0.4 ms. Consequently, loss obtained by the dual-FLC MPPT with obtained by the dual-FLC MPPT with only 2.5 µ J as compared to the single-FLC MPPT with 4.2 µ J. obtained dual-FLC MPPT with only 2.5 µ J as compared to the single-FLC MPPT with 4.2both µ J. only 2.5 µJby asthe compared to the single-FLC µJ. Interestingly, difference between Interestingly, big difference between bothMPPT MPPTswith can 4.2 be seen during boostbig operation. While the dualInterestingly, big difference between both MPPTs can be seen during boost operation. While the dualMPPTs can be seen during boost While the(energy dual-FLC 12 ms to achieve FLC MPPT only needs 12 ms to operation. achieve steady state lossMPPT of 4.8only mJ),needs the single-FLC MPPT FLC MPPT only needs 12 ms to achieve steady state (energy loss of 4.8 mJ), the single-FLC MPPT steady state to (energy of 4.8 single-FLC MPPT contributes to bigger energy loss (up to contributes biggerloss energy lossmJ), (upthe to 210 mJ) due to longer response time (100 ms). contributes loss (up(100 to 210 mJ) due to longer response time (100 ms). 210 mJ) due to to bigger longer energy response time ms). 150 150 100 100 50 50 1.3 1.3

DUAL FLC FLC DUAL FLC

1.35 1.35

1.4 1.4

1.45 1.45

1.5 1.5

1.55

Time 1.55(s) Time (a) (s) (a)

Figure 10. Cont.

1.6 1.6

1.65 1.65

1.7 1.7

FLC 1.75 1.75

1.8 1.8

Input Power(W)

Energies 2016, 9, 604 Energies 2016, 9, 604

11 of 17 11 of 17

170 160 150 140 130

DUAL FLC FLC

120 1.4995

1.5

1.5005

Time (s)

1.501

Input Power (W)

(b) 150

DUAL FLC FLC

100

50

0 0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

Time (s)

Input Power (W)

(c) 42 40 38 36 34

DUAL FLC FLC

1

1.02

1.04

1.06

1.08

1.1

1.12

Time (s) (d) Figure Figure 10. 10. Input Input powers powers under under simulation simulation obtained obtained (a) (a) from from low low to to high high irradiance irradiance or orbuck buckoperation; operation; (b) for the focused transient part in buck operation; (c) from high to low irradiance or boost (b) for the focused transient part in buck operation; (c) from high to low irradiance or boostoperation; operation; and boost operation. operation. and (d) (d) for for the the focused focused transient transient part part in in boost

Figure 11 shows output voltages at both loads 1 and 2 during buck and boost operations, Figure 11 shows output voltages at both loads 1 and 2 during buck and boost operations, respectively. During buck operation with high irradiance, there will be voltage overshoot with respectively. During buck operation with high irradiance, there will be voltage overshoot with smaller value (0.1 V) achieved by the dual-FLC MPPT as compared to the single-FLC MPPT (3.2 V). smaller value (0.1 V) achieved by the dual-FLC MPPT as compared to the single-FLC MPPT (3.2 V). Furthermore, no such oscillation occurs with the dual-FLC MPPT, while the single-FLC MPPT shows Furthermore, no such oscillation occurs with the dual-FLC MPPT, while the single-FLC MPPT shows some oscillations before reaching steady state. The negligible oscillation confirms the effectiveness of some oscillations before reaching steady state. The negligible oscillation confirms the effectiveness using direct duty cycle by the dual-FLC MPPT. During boost operation when irradiance is low, of using direct duty cycle by the dual-FLC MPPT. During boost operation when irradiance is low, negligible undershoot with faster response time (less than 0.04 s) is achieved by the dual-FLC MPPT; negligible undershoot with faster response time (less than 0.04 s) is achieved by the dual-FLC MPPT; whereas the single-FLC MPPT causes a longer time for the load 2 to achieve a stable output voltage whereas the single-FLC MPPT causes a longer time for the load 2 to achieve a stable output voltage with a response time of up to 0.17 s, with an undershoot voltage of 4.2 V. Therefore, from all with a response time of up to 0.17 s, with an undershoot voltage of 4.2 V. Therefore, from all simulation simulation results, clearly the dual-FLC MPPT shows better performance, with negligible oscillation, results, clearly the dual-FLC MPPT shows better performance, with negligible oscillation, small small overshoot, fast response time, and negligible undershoot. overshoot, fast response time, and negligible undershoot.

Energies 2016, 9, 604

12 of 17

Energies 2016, 9, 604

12 of 17 12 of 17

42 40

Voltage load 1 (V)

Voltage load 1 (V)

Energies 2016, 9,44604

38 36

34 32 30

44

42 40 38 36

34 32

0.7

0.8

30

45 40 35 30 25 20 15 10 5 0 -5

Voltage load 2 (V)

Voltage load 2 (V)

0.7

0.9 0.8

1

1.1

Time (s)1.1 (a)Time (s)

0.9

1

1.2 1.2

1.3 1.3

(a) 45 40 35 30 25 20 15 10 5 0 -5

1.3 1.3

1.41.4

1.5 1.5

1.6 1.6

Time (s) Time (s)

1.7 1.7

1.8 1.8

(b) (b) Figure 11. Output voltages of simulation work (a) at load 1 during boost operation; and (b) at load 2

Figure 11. Output voltages of simulation work (a) at load 1 during boost operation; and (b) at load 2 during buck operation. during buck operation.

5. Experimental Results

5. Experimental ExperimentalResults Results

A laboratory prototype of the PV-based SEPIC system was developed as shown in Figure 12,

A laboratory SEPIC system developed shown Figure 12, 12, with same parameters to investigate performance of the dual-FLC A Chroma Programmable prototype of the PV-based was MPPT. developed as shown in Figure Power Supply 62100H-600S (600performance V/25 A/15 kW, ATE Inc., Taoyuan, Taiwan)Programmable with Solar with DC same parameters to investigate ofChroma the dual-FLC MPPT. A Chroma Array Simulation is used as solar simulator. ThekW, dual-FLC MPPT as the proposed algorithm, and alsoSolar DC Power Supply 62100H-600S (600 V/25 A/15 kW,Chroma Chroma ATE Inc., Taoyuan, Taiwan) with V/25 A/15 ATE Inc., Taoyuan, Taiwan) the single-FLC MPPT for comparison purpose, are implemented in a TMS320F28335 eZdsp board Array Simulation Simulation isisused usedasassolar solarsimulator. simulator.The The dual-FLC MPPT as the proposed algorithm, and dual-FLC MPPT as the proposed algorithm, and also (Texas Instruments, Dallas, TX, USA). The same waveforms and parameters as defined in the also the single-FLC MPPT for comparison purpose, are implemented in a TMS320F28335 the single-FLC MPPT for comparison purpose, are implemented in a TMS320F28335 eZdsp eZdsp board simulation work are used. The measured and calculated parameters obtained for all simulation board Instruments, Dallas, TX, USA). same waveforms and parametersasasdefined defined in in the (Texas(Texas Instruments, Dallas, TX, USA). TheThe same waveforms and parameters results and experimental results are summarized in Table 4. simulation work are used. The measured and calculated parameters obtained for all simulation results simulation work are used. The measured and calculated parameters obtained for all simulation and experimental results are summarized in Tablein 4.Table 4. results and experimental results are summarized

SEPIC CONVERTER LA25NP (current sensor )

Load 1

SEPIC CONVERTER LA25NP (current sensor )

Load 1

Load 2

LV25NP (voltage sensor )

IGBT driver circuit

DSP TMS320F28335

Load changing controller

Figure 12. Laboratory prototype of the PV-based SEPIC system. Load 2 LV25NP (voltage sensor )

IGBT driver circuit

DSP TMS320F28335

Load changing controller

Figure SEPIC system. system. Figure 12. 12. Laboratory Laboratory prototype prototype of of the the PV-based PV-based SEPIC

Energies 2016, 9, 604

13 of 17

Table 4. Summary of simulation and experimental results. Energies 2016, 9, 604

MPPT

13 of 17

Simulation

Parameter

Position

Experiment

Table 4. Summary of simulation and experimental Buck results. Boost

Dual-FLC 0.1Simulation ms 12 ms MPPT Position Parameter Response Time (s) Single-FLC 0.4 ms Boost 100 ms Buck PV Input Dual-FLC 0.1 12 Dual-FLC 2.5 ms µJ 4.8ms mJ Response Time Energy Loss (J) (s) Single-FLC 0.4 ms Single-FLC 420 µJ 100 210ms mJ PV Input Dual-FLC 2.5 µ J 4.8 mJ Dual-FLC 0.015 s 0.04 s EnergyTime Loss (s) (J) Response Single-FLC 420 mJs Single-FLC 0.08µsJ 210 0.17 Dual-FLC Load 0.015 s 0.04 s Response Time (s) Dual-FLC 0.10 V ´0.1 V Overshoot/Undershoot Voltage (V) Single-FLC 0.08 s 0.17 s Single-FLC 3.2 V ´4.2 V Load Dual-FLC 0.10 V −0.1 V Overshoot/Undershoot Voltage (V) Single-FLC 3.2 V −4.2 V

Buck

Boost

Experiment 0.01 s 0.02 s 0.2 s Boost 0.26 s Buck 0.01 s 0.02 s mJ 4 mJ 19 0.2 s mJ 860s mJ 0.26293 4 mJ 19 mJ 0.03 s 0.01 s 8600.12 mJ s 293 mJ 0.3 s 0.03 s 0.01 s 0.01 V ´0.35 V 0.12 s 0.3 s 5.3 V ´6.8 V 0.01 V −0.35 V 5.3 V −6.8 V

Figure 13 shows waveforms of input powers for both buck and boost operations. During buck 13 the shows waveformswork, of input for both buckperforms and boostbetter operations. buck time operation,Figure like in simulation thepowers dual-FLC MPPT with During a response operation, like in the work, the dual-FLC better with a response time of by of 0.01 s as compared tosimulation the single-FLC MPPT with MPPT 0.2 s. performs Reduction of energy loss is obtained 0.01 s as compared to the single-FLC MPPT with 0.2 s. Reduction of energy loss is obtained by the the dual-FLC MPPT (only 4 mJ) as compared to the single-FLC MPPT with 860 mJ. During boost, dual-FLC MPPT (only 4 mJ) as compared to the single-FLC MPPT with 860 mJ. During boost, a a significant impact of implementing the dual-FLC MPPT is shown, with only 0.02 s needed to achieve significant impact of implementing the dual-FLC MPPT is shown, with only 0.02 s needed to achieve steady state (lower energy loss of 19 mJ), while the single-FLC MPPT causes a bigger energy loss (up to steady state (lower energy loss of 19 mJ), while the single-FLC MPPT causes a bigger energy loss 293 mJ) to mJ) the due longer response time (0.26 s).(0.26 s). (updue to 293 to the longer response time

(a) Input Power (W)

200 Dual FLC FLC

150 100 50 0 8

9

10

11

12

13

14

Time (s) Input Power (W)

(b) Dual FLC FLC

80 60 40 20 0

9

9.5

10

Time (s)

10.5

11

11.5

(c) Figure 13. Input powers under experimental work obtained (a) from low to high irradiance or buck

Figure 13. Input powers under experimental work obtained (a) from low to high irradiance or buck operation; (b) from high to low irradiance or boost operation; and (c) for the focused transient part in operation; (b) from high to low irradiance or boost operation; and (c) for the focused transient part in boost operation. boost operation.

Energies 2016, 9, 604 Energies 2016, 9, 604

14 of 17 14 of 17

Voltage load 1 (V)

Figure shows output Figure 14 14 shows output voltages voltages at at both both loads loads 11 and and 22 during during buck buck and and boost boost operations, operations, respectively. voltage with thethe dual-FLC MPPT as respectively. During Duringbuck buckoperation, operation,there therewill willbebenonoovershoot overshoot voltage with dual-FLC MPPT compared to the single-FLC MPPT (5.3 V). During boost operation, lower undershoot with faster as compared to the single-FLC MPPT (5.3 V). During boost operation, lower undershoot with faster response time(less (lessthan than0.01 0.01s)s) achieved dual-FLC MPPT; however, the single-FLC response time is is achieved by by thethe dual-FLC MPPT; however, the single-FLC MPPTMPPT takes takes (a response to load 0.3 s)2 for load 2a stable to obtain a stable output voltage. All longerlonger (a response time of time up toof 0.3up s) for to obtain output voltage. All experimental experimental results confirm the of effectiveness of MPPT the dual-FLC MPPT in SEPIC operation to step results confirm the effectiveness the dual-FLC in SEPIC operation to step down and up down and up voltage accordingly. voltage accordingly.

45 40 35 30

Dual FLC FLC

25 9.5

10

10.5

Time (s)

11

11.5

12

(a) Voltage Load 2 (V)

22

Dual FLC FLC

20 18 16 4.8

5

5.2

5.4

5.6

Time (s) (b)

5.8

6

6.2

6.4

Figure 14. Output voltages of experimental work (a) at load 1 during boost operation; and (b) at load Figure 14. Output voltages of experimental work (a) at load 1 during boost operation; and (b) at load 2 2 during buck operation. during buck operation.

Changing the single-FLC MPPT to single-FLC 1 MPPT and single-FLC 2 MPPT, the performance thebeen single-FLC MPPT single-FLC 1 MPPT and single-FLC 2 MPPT, performance of theChanging MPPTs has evaluated andtothis case study is rather important to justify thethe selection of the of the MPPTs has been evaluated and this case study is rather important to justify the selection the membership function pattern when applying FLC MPPT. Figure 15 shows the output voltages atofboth membership function pattern when applying FLC MPPT. Figure 15 shows the output voltages at both loads 1 and 2 during buck and boost operations by comparing the performance of single-FLC 1 with loadsFLC 1 and 2 during buckboost and operation, boost operations by comparing the single-FLC performance of single-FLC 1 with dual MPPT. During the performance of both 1 MPPT and dual-FLC dual FLC MPPT. During boost operation, the performance of both single-FLC 1 MPPT and dual-FLC MPPT are the same due to their similar membership function patterns. During buck operation, there MPPT same due to their similar membership function patterns.toDuring buck operation, will beare no the overshoot voltage with the dual-FLC MPPT as compared the single-FLC 1 MPPTthere (5.3 will be no overshoot voltage with the dual-FLC MPPT as compared to the single-FLC 1 MPPT (5.3 V). V). By using single-FLC 1, there is no undershoot voltage during boost operation but overshoot By usingstill single-FLC 1, there is no undershoot voltage during boost operation but overshoot voltage voltage exists during buck operation. still exists buckoutput operation. Figureduring 16 shows voltages at both loads 1 and 2 during buck and boost operations by Figure 16 shows output at both loads 1 and 2 MPPT. during During buck and boost operations comparing the performance ofvoltages single-FLC 2 with dual-FLC buck operation, the by comparing the performance of single-FLC 2 with dual-FLC MPPT. During buck operation, performance of both single-FLC 2 MPPT and dual-FLC MPPT are the same due to their similar the performance of both single-FLC 2 MPPT and dual-FLC same due to their similar membership function patterns. During boost operation, thereMPPT will beare no the undershoot voltage obtained membership function patterns. During boost operation, there will be no undershoot voltage obtained by the dual-FLC MPPT as compared to the single-FLC 2 MPPT. Single-FLC 2 MPPT also presents a by the dual-FLC MPPT compared the single-FLC MPPT also presents greater time response of as about 0.2 s. Bytousing single-FLC22,MPPT. there isSingle-FLC undershoot2 voltage during boost a greater time response of about 0.2 s. By using single-FLC 2, there is undershoot voltage operation but no overshoot voltage exists during buck operation. Table 5 presents simplified during results boostshows operation no overshoot voltagevoltage exists during buck operation. Table 5 presents simplified that the but occurrence of overshoot and undershoot voltage during buck and boost results that shows the occurrence of overshoot voltage and undershoot voltage during buck and condition for each MPPT algorithm and it clearly shows the need of dual FLC MPPT to buckboost and condition for each MPPT algorithm and it clearly shows the the needinitial of dual FLC MPPT to buck and boost boost output voltage without having energy losses during period. output voltage without having energy losses during the initial period.

Energies 2016, 9, 604 Energies 2016, 9, 604 Energies 2016, 9, 604

15 of 17 15 of 17 15 of 17

V Voltage o lta g e loload a d 1 1(V(V) )

50 50 45 45 40 40 35 35 30 30

Dual FLC Dual FLC Single FLC 1 Single FLC 1

25 25 20 20 9.5 9.5

10 10

10.5 10.5

11 11

Time (s) Time (s)

11.5 11.5

12 12

(a) (a)

V Voltage o lta g e L oLoad a d 2 2(V(V) )

22 22

Dual FLC Dual FLC Single FLC 1 Single FLC 1

20 20 18 18 16 16

4.8 4.8

5 5

5.2 5.2

5.4 5.4

5.6

5.8 5.8

5.6(s) Time Time (s)

6 6

6.2 6.2

6.4 6.4

(b) (b)

Figure 15. 15. Output voltages of of experimentalwork work (comparingDual Dual FLCwith with SingleFLC FLC 1), (a) at load1 Figure Output voltages Figure 15. Output voltages of experimental experimental work(comparing (comparing DualFLC FLC withSingle Single FLC1),1),(a)(a)atatload load 1 during boost operation; and (b) at load 2 during buck operation. during boost operation; and (b)(b) at at load 2 during buck operation. 1 during boost operation; and load 2 during buck operation.

Voltage load 1 (V)

Voltage load 1 (V)

45

45 40

40

35

35 30

30

Dual FLC Single FLC 2

25 9.5

10

25 9.5

10.5

10

11

Time (s) (a)

10.5

11.5

11

Dual FLC Single12 FLC 2

11.5

12

Time (s) Voltage V o ltage lo ad load 2 (V )2 (V)

Dual FLC Single FLC 2 Dual FLC Single FLC 2

(a)

19 18.5 19 18 18.5

17.5 18 17 17.5

16.5 17 16 16.5

15.5 16 4.8

5

5.2

5.4

5.6

5.8

6

6.2

5.4

Time (s) 5.6 (b) Time (s)

5.8

6

6.2

15.5 4.8

5

5.2

Figure 16. Output voltages of experimental work (b) (comparing Dual FLC with Single FLC 2), (a) at load 1 during boost operation; and (b) at load 2 during buck operation. Figure 16. 16. Output voltages of of experimental experimentalwork work(comparing (comparingDual DualFLC FLCwith withSingle SingleFLC FLC2),2),(a)(a)atatload load1 Figure Output voltages 1 during boost operation; and load 2 during buck operation. during boost operation; and (b)(b) at at load 2 during buck operation.

Energies 2016, 9, 604

16 of 17

Table 5. Occurrence of overshoot and undershoot voltage during buck and boost with MPPT algorithms. MPPT Algorithms

Overshoot Voltage Occur during Buck Operation

Undershoot Voltage Occur during Boost Operation

Dual FLC (FLC 1 and FLC 2) Single FLC Single FLC 1 Single FLC 2

No Yes Yes No

No Yes No Yes

6. Conclusions This paper has successfully presented significant work on dual-FLC MPPT for SEPIC. The results obtained, from both simulation and experimental work, clearly show the effectiveness of the dual-FLC MPPT as compared to the single-FLC MPPT. In the dual-FLC MPPT, both FLC parts, assigned for buck and boost operations, significantly overcome the problems of overshoot and undershoot voltages. Overcoming these problems consequently contributes to a reduction of energy losses. In addition, besides addressing MPPT for obtaining the maximum power in steady state operation, the proposed MPPT has successfully proven to function in situations of fast changing irradiance. Acknowledgments: This work is supported by the Fundamental Research Gant Scheme under the Ministry of Higher Education, Malaysia (03-01-14-1413FR). Author Contributions: Tanaselan Ramalu designed and developed the main parts of the research work, including simulation model, experimental set up, and analyses of the obtained results. Tanaselan Ramalu was also mainly responsible for preparing the paper. Mohd Amran Mohd Radzi contributed in simulation, experimental, and writing parts. Muhammad Ammirrul Atiqi Mohd Zainuri, Noor Izzri Abdul Wahab and Ribhan Zafira Abdul Rahman also involved in verifying the work and actively contributed to finalize the manuscript. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4. 5.

6.

7.

8.

9. 10.

Parida, B.; Iniya, S.; Goic, R. A review of solar photovoltaic technologies. Renew. Sustain. Energy Rev. 2011, 15, 1625–1636. [CrossRef] Ding, K.; Bian, X.G.; Liu, H.H.; Peng, T. A MATLAB-simulink-based PV module model and its application under conditions of nonuniform irradiance. IEEE Trans. Energy Convers. 2012, 27, 864–872. [CrossRef] Krismadinataa; Rahim, N.A.; Pinga, H.W.; Selvaraja, J. Photovoltaic module modeling using simulink/matlab. Procedia Environ. Sci. 2013, 17, 537–546. [CrossRef] Qi, C.; Ming, Z. Photovoltaic module simulink model for a stand-alone PV system. Phys. Procedia 2012, 24, 94–100. [CrossRef] Durán, E.; Cardona, M.S.D.; Galán, J.; Andújar, J.M. Comparative Analysis of Buck-Boost Converters Used to Obtain I-V Characteristic Curves of Photovoltaic Modules. In Proceedings of the Power Electronics Specialists Conference, Rhodes, Greece, 15–19 June 2008. Mohd Zainuri, M.; Radzi, M.A.M.; Soh, A.C.; Rahim, N.A. Development of Adaptive Perturb and Observe-Fuzzy Control Maximum Power Point Tracking for Photovoltaic Boost DC-DC Converter. IET Renew. Power Gener. 2014, 8, 183–194. [CrossRef] Taghvaee, M.H.; Radzi, M.A.M.; Moosavain, S.M.; Hizam, H.; Marhaban, M.H. A current and future study on non-isolated DC–DC converters for photovoltaic applications. Renew. Sustain. Energy Rev. 2013, 17, 216–227. [CrossRef] El Khateba, A.H.; Rahim, N.A.; Selvaraja, J. Fuzzy Logic Control Approach of a Maximum Power Point Employing SEPIC Converter for Standalone Photovoltaic System. In Proceedings of the 3rd International Conference on Sustainable Future for Human Security, Kyoto, Japan, 3–5 November 2012. Khateb, A.E.; Rahim, N.A.; Selvaraj, J.; Uddin, M.N. Fuzzy-Logic-Controller-Based SEPIC Converter for Maximum Power Point Tracking. IEEE Trans. Ind. Appl. 2014, 50, 2349–2358. [CrossRef] Salas, V.; Olías, E.; Barrado, A.; Lázaro, A. Review of the Maximum Power Point Tracking Algorithms for Stand-Alone Photovoltaic Systems. Sol. Energy Mater. Sol. Cells 2006, 90, 1555–1578. [CrossRef]

Energies 2016, 9, 604

11. 12. 13. 14. 15. 16.

17. 18. 19. 20. 21. 22.

23.

17 of 17

Liu, C.-L.; Chen, J.-H.; Liu, Y.-H.; Yang, Z.-Z. An Asymmetrical Fuzzy-Logic-Control-Based MPPT Algorithm for Photovoltaic Systems. Energies 2014, 7, 2177–2193. [CrossRef] Cheng, P.-C.; Peng, B.-R.; Liu, Y.-H.; Cheng, Y.-S.; Huang, J.-W. Optimization of a Fuzzy-Logic-Control-Based MPPT Algorithm Using the Particle Swarm Optimization Technique. Energies 2015, 8, 5338–5360. [CrossRef] Lalounia, S.; Rekioua, D.; Rekioua, T.; Matagne, E. Fuzzy Logic Control of Stand-Alone Photovoltaic System with Battery Storage. J. Power Sources 2009, 193, 899–907. [CrossRef] Salam, Z.; Ahmed, J.; Merugu, B.S. The application of soft computing methods for MPPT of PV system: A technological and status review. Appl. Energy 2013, 107, 135–148. [CrossRef] Nabulsi, A.A.; Dhaouadi, R. Efficiency Optimization of a DSP-Based Standalone PV System Using Fuzzy Logic and Dual-MPPT Control. IEEE Trans. Ind. Inform. 2012, 8, 573–584. [CrossRef] Cakmak, R.; Altas, I.H.; Sharaf, A.M. Modeling of FLC-Incremental Based MPPT Using DC-DC Boost Converter for Standalone PV System. In Proceedings of the 2012 International Symposium Innovations in Intelligent Systems and Applications (INISTA), Trabzon, Turkey, 2–4 July 2012. Altasa, I.H.; Sharafb, A.M. A Novel Maximum Power Fuzzy Logic Controller for Photovoltaic Solar Energy Systems. Renew. Energy 2008, 33, 388–399. [CrossRef] Karthika, S.; Rathika, P.; Devaraj, D. Fuzzy logic based maximum power point tracking designed for 10 kW solar photovoltaic system. Int. J. Comput. Sci. Manag. Res. 2013, 2, 1421–1427. Rahmani, R.; Seyedmahmoudian, M.M.; Mekhilef, S.; Yusof, R. Implementation of Fuzzy Logic Maximum Power Point Tracking Controller for Photovoltaic System. Am. J. Appl. Sci. 2013, 10, 209–218. [CrossRef] Salah, C.B.; Ouali, M. Comparison of fuzzy logic and neural network in maximum power point tracker for PV systems. Electr. Power Syst. Res. 2011, 81, 43–50. [CrossRef] Chiang, S.J.; Shieh, H.J.; Chen, M.C. Modeling and Control of PV Charger System with SEPIC Converter. IEEE Trans. Ind. Electron. 2009, 56, 4344–4353. [CrossRef] Hajighorbani, S.; Radzi, M.A.M.; Kadir, M.Z.A.A.; Shafie, S. Dual Search Maximum Power Point (DSMPP) Algorithm Based on Mathematical Analysis under Shaded Conditions. Energies 2015, 8, 12116–12146. [CrossRef] Bo, F.; Lu, Y.; Qiong, C.; Xinchen, C.; Xin, X. A Photovoltaic MPPT Strategy by Modified Fuzzy Control. Appl. Mech. Mater. 2012, 128–129, 15–19. © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).

A Photovoltaic-Based SEPIC Converter with Dual-Fuzzy Maximum Power Point Tracking for Optimal Buck and Boost Operations Tanaselan Ramalu 1,2, *, Mohd Amran Mohd Radzi 1,2 , Muhammad Ammirrul Atiqi Mohd Zainuri 1,2 , Noor Izzri Abdul Wahab 1,2 and Ribhan Zafira Abdul Rahman 1 1

2

*

Department of Electrical and Electronic Engineering, Faculty of Engineering, University Putra Malaysia, 43400 Serdang, Selangor, Malaysia; [email protected] (M.A.M.R.); [email protected] (M.A.A.M.Z.); [email protected] (N.I.A.W.); [email protected] (R.Z.A.R.) Centre for Advanced Power and Energy Research, Faculty of Engineering, University Putra Malaysia, 43400 Serdang, Selangor, Malaysia Correspondence: [email protected]; Tel.: +60-17-2566-574

Academic Editor: Gabriele Grandi Received: 17 May 2016; Accepted: 27 July 2016; Published: 30 July 2016

Abstract: In this paper, a photovoltaic (PV)-based single ended primary-inductor converter (SEPIC) is developed with introduction of dual-fuzzy logic controller (FLC) maximum power point tracking (MPPT) algorithm. Separate FLC parts, for the first time used for MPPT, are configured for optimal operations of both buck and boost operations. During buck operation, a high overshoot voltage exists, and during boost operation, an undershoot voltage occurs, both during the initial rising period. Definitely, a single-FLC MPPT could not be able to minimize both problems, which on the other hand can be handled by the proposed MPPT algorithm. For evaluation purposes, buck operation has been conducted during high irradiance, while during low irradiance, boost operation has been conducted. The dual-FLC MPPT with SEPIC was simulated in MATLAB-Simulink, and further a laboratory prototype was implemented with a TMS320F28335 eZdsp board. Both simulation and experimental results and comparison analysis (with the single-FLC MPPT) have been presented. From the results and analysis, the dual-FLC MPPT performs better than the single-FLC MPPT in terms of faster response time, lower overshoot and undershoot, and further significant reduction of power losses. Keywords: photovoltaic (PV); maximum power point tracking (MPPT); fuzzy logic controller (FLC); single ended primary-inductor converter (SEPIC)

1. Introduction In terms of renewable energy, there is a long list of energy types that comes from various natural resources such as solar, wind, geothermal, sea tide, and biomass. Among them, photovoltaic (PV) from solar is much preferable due to its implementation simplicity with less maintenance. In recent years, PV systems have witnessed neverending demand due to their enormous potential to be the nearest solution we have right now to substitute our diminishing fossil fuel energy sources. When a PV panel is exposed to solar irradiation, it can generate direct current electricity without any environmental impact or contamination. The only drawbacks are that the cost to manufacture PV panels is too high and their small range of efficiency is only about 15%–20% [1–3]. PV panels have nonlinear output characteristics and the main factors affecting PV output power are the solar irradiation, temperature and load impedance [3]. When the solar irradiation rises, the PV current increases, however, the temperature of a PV module has a more significant effect on PV voltage operation [4]. Due to the nonlinear output characteristics of PV panels, an algorithm is needed to track Energies 2016, 9, 604; doi:10.3390/en9080604

www.mdpi.com/journal/energies

Energies 2016, 9, 604

2 of 17

the maximum power point (MPP) of the PV curve to deliver the maximum power. This is known as maximum power point tracking (MPPT). The MPPT operation basically involves finding the maximum operating voltage and current at which PV operates to achieve the MPP. Many MPPT methods have been developed and implemented [4,5]. Among them are perturb and observe (P&O), incremental conductance (IC), artificial neural network (ANN), fuzzy logic controller (FLC), constant voltage, three-point weight comparison, short current pulse, and open circuit voltage. The most commonly used traditional methods are P&O and incremental conductance, however, potential artificial intelligent techniques like FLC and ANN are recently gaining popularity in MPPT design due to their ability to achieve higher stability and less noise factor [6]. Specifically, FLC does not require an accurate mathematical model and is known to be very efficient in handling problems that have non-linear variables [6]. Meanwhile, the amount of DC-based equipment operated at various levels is growing higher, and thus, having a DC-DC converter with ability to produce various DC outputs is preferable. The SEPIC converter is preferable due to its ability to buck and boost input voltage and has advantage of having non-inverted output. By having non-inverted output polarity, implementation of circuit to load becomes easier due to the fact the reference point (ground) is the same. There are quiet a number of significant works on SEPIC with PV [7], which also covers MPPT algorithms, including FLC [8]. As MPPT tracks maximum power via increasing or decreasing voltage and current, SEPIC is effective as it increases and decreases voltage at the current’s expense [9]. However, in related to FLC MPPT for SEPIC, there are no such works considering comprehensive evaluations of MPPT for both buck and boost operations. By using only a single FLC for MPPT, although it performs better as compared to P&O [10], using just a common pattern of membership functions may degrade its performance to track MPP, especially when facing dynamic irradiance changes. In addition, at the controller output for producing a PWM signal to power switching devices in SEPIC, this type of MPPT is only suitable to be set as a duty cycle change which will be added to a pre-defined duty cycle. Use of a direct duty cycle approach is totally unsuitable. Consequently, when the irradiance changes rapidly, there is a possibility that the single-FLC MPPT will fail to track its MPP and take a certain amount of time to reach steady state conditions [11,12]. Therefore, this paper proposes a dual-FLC MPPT that offers a significant improvement of both buck and boost operations in SEPIC. Design of membership functions for each FLC will be carried out by focusing to each specific problem in buck and boost operations, respectively. Hypothetically, as further proven later in this paper, during buck operation, a high overshoot voltage exists, and during boost operation, an undershoot voltage occurs, both during the initial period of changes. Thus, the proposed MPPT should ensure not only operation of SEPIC at MPP, but also should address the mentioned problems, both in simulation and experimental works. Two separate loads with a switching circuit, called as load changing circuit, are set up to ensure a significant impact of dynamic changes can be provided during evaluation of the proposed MPPT. As for the rest of this paper, Section 2 covers the proposed PV-based system, followed by a description of the proposed MPPT in Section 3, and both simulation and experimental validations in Sections 4 and 5, respectively. Finally Section 6 concludes the findings. 2. Proposed Photovoltaic-based SEPIC System The proposed PV-based SEPIC system, as shown in Figure 1, consists of five main parts: PV panel, SEPIC, controller which consists of dual-FLC MPPT and load changing algorithms, load changing circuit, and loads. Meanwhile, Figure 2 presents the configuration of SEPIC. Voltage and current are measured and used by the dual-FLC MPPT to produce a suitable duty cycle to operate IGBT in SEPIC.

Energies 2016, 9, 604 Energies 2016, 9, 604

3 of 17 3 of 17

Load changing circuit

Sepic converter

PV Panel

Load 1

Duty cycle V

I DUAL FLC MPPT

Load 2

Load changing algorithm DSP

Figure proposedPV-based PV-basedSEPIC SEPICsystem. system. Figure1.1.Block Block diagram diagram of of proposed Inductor 1

Capacitor 1

Diode

PV panel Input Capacitor

IGBT

Inductor 2 Capacitor 2

Load

Figure2.2.Configuration Configuration of Figure of SEPIC SEPIC with withPV PVasasinput. input.

The voltage conversion ratio of SEPIC can be defined as follows: The voltage conversion ratio of SEPIC can be defined as follows:

𝐷 𝑉𝑜𝑢𝑡 =ˆ( D ˙) 𝑉𝑖𝑛 Vout “ 1 − 𝐷 Vin

(1)

(1) 1´D where Vout and Vin are the output and input voltages of SEPIC, respectively, and D is the duty cycle which is defined thethe ratio of theand turn-on the switching time period. where Vout and Vinas are output inputduration voltagestoof SEPIC, respectively, and D is the duty cycle Two separate loads are connected through a load changing circuit, considering one load is which is defined as the ratio of the turn-on duration to the switching time period. specifically used for buck operation and another one for boost operation. Critical performance during Two separate loads are connected through a load changing circuit, considering one load is the switching between buck boost one operations be investigated A threshold specifically used period for buck operation andand another for boostcan operation. Critical later. performance during current is set as 3 A for the load changing algorithm. If the current is less than the threshold value, the switching period between buck and boost operations can be investigated later. A threshold current the as circuit willthe connect SEPIC to load 1. OnceIfthe current is more thanthreshold the threshold value, the is set 3 A for load changing algorithm. theinput current is less than the value, the circuit circuit will connect SEPIC to load 2. When irradiance is low, the current delivered from the PV panel will connect SEPIC to load 1. Once the input current is more than the threshold value, the circuit will will be lower and SEPIC performs boost operation with FLC 1. When the irradiance is higher, the connect SEPIC to load 2. When irradiance is low, the current delivered from the PV panel will be lower input current from the PV panel will increase, so SEPIC will perform buck operation with FLC 2. and SEPIC performs boost operation with FLC 1. When the irradiance is higher, the input current from the3.PV panel willLogic increase, so SEPIC will perform Dual-Fuzzy Controller Maximum Powerbuck Pointoperation Tracking with FLC 2. As mentioned, FLCs areMaximum used as MPPT in this system to perform buck and boost operations, 3. Dual-Fuzzy Logictwo Controller Power Point Tracking respectively. The same two inputs (error E and change of error CE) at sample time k are used, which As mentioned, two FLCs are used as MPPT in this system to perform buck and boost operations, are defined as below: respectively. The same two inputs (error E and change of error CE) at sample time k are used, which 𝑃(𝑘)−𝑃(𝑘−1) are defined as below: (2) E (k) = P pkq ´ P pk ´ 1q E pkq “ 𝑉(𝑘)−𝑉(𝑘−1) (2) V pkq ´ V pk ´ 1q CE (k) = E (k) − E (k − 1) (3) CE pkq “ E pkq ´ E pk ´ 1q (3) where E is the change of PV power over the change of PV voltage, and CE is the difference between where E is theEchange PV power the sample change time. of PV voltage, and CE is the difference between the current from theofprevious E atover a given Basically, thethe operation of E FLC be classified into four main elements: fuzzification, rule base, the current E from previous at acan given sample time. inference engine defuzzification [13–17]. During fuzzification, the inputs of thefuzzification, FLC, CE and rule E Basically, the and operation of FLC can be classified into four main elements: variables are calculated anddefuzzification converted into linguistic based on the membership base, inference engine and [13–17]. variables During fuzzification, the inputs functions. of the FLC, The output (in this case it is the duty cycle D) is generated by looking up in a rule base table [18]. The CE and E variables are calculated and converted into linguistic variables based on the membership fuzzy output is converted back to a numerical variable from a linguistic variable during functions. The output (in this case it is the duty cycle D) is generated by looking up in a rule base

Energies 2016, 9, 604 Energies 2016, 9, 604

4 of 17 4 of 17

Energies 2016, 9, 604

4 of 17

table [18]. The fuzzy output converted backFLCs to a numerical from a linguistic variable defuzzification [19–23]. Theisdesign of both differs invariable their patterns and ranges, as to during reflect defuzzification [19–23]. The design of both FLCs differs in their patterns and ranges, as to reflect effectiveness of buck and boost operations, respectively. The concept of designing them is based defuzzification [19–23]. The design of both FLCs differs in their patterns and ranges, as to reflect on effectiveness buck and boost operations, respectively. The designing them based the PV curve,ofand mapping ofoperations, membership functions inconcept the PVof curve for the proposed MPPT is effectiveness of the buck and boost respectively. The concept of designing them isisbased onon the thein PVFigure curve, and the mapping of membership functions thecurve PV curve for the proposed MPPT shown PV curve, and the3. mapping of membership functions in theinPV for the proposed MPPT is is shown shown in Figure 3. in Figure 3. Power(P) Power(P)

Voltage(V)

Voltage(V) Figure 3. Mapping of the membership functions in the P-V curve for dual-FLC MPPT.

Figure 3. Mapping of the membership functions in the P-V curve for dual-FLC MPPT. Figure 3. Mapping of the membership functions in the P-V curve for dual-FLC MPPT. The PV curve has been divided into seven segments with each representing different The hashas beenin divided into seven with eachwith representing different membership membership functions the dual-FLC MPPT. The seven membership functions are Negative Big The PV PVcurve curve been divided into segments seven segments each representing different (NB), Negative Medium (NM), Negative Small (NS), Zero (ZE), Positive Small (PS), Positive Medium functions in the dual-FLC MPPT. The seven membership functions are Negative Big (NB), Negative membership functions in the dual-FLC MPPT. The seven membership functions are Negative Big and Positive Big (NM), (PB). ZE membership function is located exactly at the maximum power point Medium (NM), Negative Small (NS), Zero (ZE), Positive Small (PS), Positive Medium (PM) and (NB),(PM) Negative Medium Negative Small (NS), Zero (ZE), Positive Small (PS), Positive Medium of the PV curve. The membership functions of the left hand side of ZE (PS, PM, PB), will be labelled Positive (PB). ZE function function is located the maximum power point of the (PM) andBig Positive Big membership (PB). ZE membership is exactly located at exactly at the maximum power point as positive polarity as the gradient of the PV curve is positive, and the membership functions of the PV curve. The membership functions of the left hand side of ZE (PS, PM, PB), will be labelled as of the PV curve. The membership functions of the left hand side of ZE (PS, PM, PB), will be labelled right hand side of ZE (NS, NM, NB) will be labelled as negative polarity as the gradient of the PV positive polarity as the gradient of the is positive, and the functions of theofright as positive polarity as the gradient of PV the curve PV curve is positive, andmembership the membership functions the curve is negative. The areas of PS and NS for both polarities, in which the controller becomes more hand side of ZE (NS, NM, NB) will be labelled as negative polarity as the gradient of the PV curve is rightsensitive hand side of ZEZE(NS, NM, NB) willduty be labelled as controller, negative are polarity as the gradient of the PV towards which determines cycle of the critical. negative. The areasThe of PS both in which thewhich controller becomes sensitive curve is negative. areas ofNS PSfor and NS polarities, for both polarities, in controller becomes As mentioned, theand design of membership functions for both FLCs the is different, as more shown in more towards ZE which determines duty cycle of the controller, are critical. sensitive towards ZE which determines duty cycle of the controller, are critical. Figures 4 and 5, respectively. FLC 1 is specially designed to overcome undershoot voltage problems As the design of membership functions voltage for both FLCs is In different, as seven shown in and mentioned, FLC 2 is specially designed to overcome overshoot problems. each FLC, membership functions areFLC configured for all inputs and output. All theundershoot membership functions are set and Figures 44and 1 is specially designed to overcome voltage problems Figures and5,5,respectively. respectively. FLC 1 is specially designed to overcome undershoot voltage problems shapes with both ending sides of the universe disclosure accompanied a trapezium FLC istriangular specially designed to overcome overshoot voltageofproblems. each FLC,Inby seven membership and 2as FLC 2 is specially designed to overcome overshoot voltage In problems. each FLC, seven shapeare to show continuous operation controller. The selected fuzzy is Mamdani functions configured forconfigured all inputs of and output. the membership functions are functions set aswhere triangular membership functions are forthe all inputsAll and output. All the method membership are set the maximum of the minimum composition technique for the inference is used. The center-of-gravity shapes with both ending the universe disclosure a trapeziumby shape to show as triangular shapes withsides bothof ending sides ofofthe universeaccompanied of disclosureby accompanied a trapezium method is used in the defuzzification process.

Degree of membership

Degree of membership

continuous operation of theoperation controller.ofThe fuzzy is fuzzy Mamdani where the maximum of shape to show continuous theselected controller. Themethod selected method is Mamdani where the technique for the inference used. center-of-gravity is used in the minimum maximumcomposition of the minimum composition techniqueisfor the The inference is used. Themethod center-of-gravity NB NM NS ZE PS PM PB the defuzzification process. method is used1 in the defuzzification process. NB

1 0.5

0.5

NM

NS ZE

PS

PM

PB

0

-2.0

-1.5

-1

-0.5

0

0.5

1

1.5

2

Error, E (ΔP/ΔV)

(a) 0

-2.0

-1.5

-1

-0.5

0

Error, E (ΔP/ΔV)

(a) Figure 4. Cont.

0.5

1

1.5

2

membership Degree membership ofof Degree

Energies 2016, 9, 604 Energies 2016, 9, 604 Energies 2016, 9, 604

5 of 17 5 of 17 5 of 17 NB NB

1 1

NM NM

NS ZE NS ZE

PS PS

PM PM

PB PB

0.5 0.5

0 0 -2.0 -2.0

-1.5 -1.5

-1 -1

-0.5 -0.5

0 0

0.5 0.5

Change of Error, CE (ΔE) Change of Error, CE (ΔE)

1 1

1.5 1.5

2 2

membership Degreeofofmembership Degree

(b) (b)

NB NB

1 1

NM NM

NS NS

ZE ZE

PS PS

PM PM

PB PB

0.5 0.5

0 00.2

0.3 0.3

0.2

0.4 0.4

0.5 0.5

0.6 0.6

Output, ΔD Output, ΔD

0.7 0.7

0.8 0.8

0.9 0.9

1 1

(c) (c)

membership Degree membership ofof Degree

Figure 4. Fuzzy membership functions of inputs and output for FLC 1: (a) error; (b) change of error; Figure functions of of inputs inputs and and output output for for FLC FLC 1: 1: (a) (a) error; error; (b) (b) change change of of error; error; Figure 4. 4. Fuzzy Fuzzy membership membership functions and (c) output. and (c) output. and (c) output. 1 1

NB NB

NM NM

NS ZE NS ZE

PS PM PS PM

PB PB

0.5 0.5

0 0-2.0

-1.5 -1.5

-2.0

-1 -1

-0.5 -0.5

0 0

Error, E (ΔP/ΔV) Error, E (ΔP/ΔV)

0.5 0.5

1 1

1.5 1.5

2 2

1.5 1.5

2 2

membership Degree membership ofof Degree

(a) (a)

1 1

NB NB

NM NM

NS ZE NS ZE

PS PM PS PM

PB PB

0.5 0.5

0 0-2.0

-2.0

-1.5 -1.5

-1 -1

-0.5 -0.5

0 0

0.5 0.5

Change of Error, CE (ΔE) Change of Error, CE (ΔE) (b) (b) Figure 5. Cont.

1 1

Energies 2016, 9, 604 Energies 2016, 9, 604

6 of 17 6 of 17

Degree of membership

NB

NM

NS

ZE

PS

PM

PB

1

0.5

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Output, ΔD

(c) Figure Figure 5. 5. Fuzzy Fuzzy membership membership functions functions of of inputs inputs and and output output for for FLC FLC 2: 2: (a) (a) error; error; (b) (b) change change of of error; error; and (c) output of FLC 2. and (c) output of FLC 2.

In FLC 1, the negative polarity membership functions such as NB, NM and NS are arranged In FLC 1, the negative polarity membership functions such as NB, NM and NS are arranged closely to ZE so that the right hand side of the PV curve is given more priority because the selected closely to ZE so that the right hand side of the PV curve is given more priority because the selected area covered under the curve has higher voltage and this could compensate the undershoot voltage. area covered under the curve has higher voltage and this could compensate the undershoot voltage. Meanwhile, in FLC 2, the positive polarity membership functions such as PB, PM and PS are arranged Meanwhile, in FLC 2, the positive polarity membership functions such as PB, PM and PS are arranged closely to ZE so that the left hand side of the PV curve is given more priority because the selected closely to ZE so that the left hand side of the PV curve is given more priority because the selected area area covered under the curve has lesser voltage and this could reduce the overshoot voltage. covered under the curve has lesser voltage and this could reduce the overshoot voltage. Switching Switching signals generated by SEPIC are based on direct duty, so as to overcome the limitation of signals generated by SEPIC are based on direct duty, so as to overcome the limitation of the single the single FLC mentioned early. However, an initial D has separately been set for FLC 1 and FLC 2 FLC mentioned early. However, an initial D has separately been set for FLC 1 and FLC 2 patterns patterns respectively for preventing the output voltage to have unwanted overshoot or undershoot respectively for preventing the output voltage to have unwanted overshoot or undershoot which could which could harm the system. Table 1 shows the 49 rules for both FLCs. harm the system. Table 1 shows the 49 rules for both FLCs. Table 1. Rules used for both FLCs. Table 1. Rules used for both FLCs.

E/CE E/CE NB NM NB NM NS NS ZE ZE PS PS PM PM PB PB

NB NB ZE ZE ZE ZE ZE ZE PS PS NS NS NM NM NB NB

NM NS ZE NM NS ZE ZE ZE PB ZE ZE PM ZE ZE PB ZE ZE ZE PM PS ZE ZE PS ZE ZE ZE ZE ZE ZE NS NS NS NS NS NS NM NM NM NM NM NM NB NB NB NB NB NB

PS

PM

PB

PS PB

PM PB

PB PB

PM PB PM PB PM PB PM PS PM PS PM PS PS PS PS ZE ZE NS ZE ZE NS ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE ZE

Finally, full operation operation of of the the controller controller is is shown shown in in Figure Figure 6. 6. After Finally, full After measuring measuring the the PV PV voltage voltage and and current, current, power power will will be be calculated. calculated. Then, Then, changes changes of of PV PV power power and and voltage voltage are are calculated. calculated. After After that, that, error error E E and and change change of of error error CE CE are are determined determined for for later later use use as as inputs inputs for forthe theselected selectedFLC. FLC.Meanwhile, Meanwhile, FLC and load must be chosen, and this process considers measurement of input current. As FLC and load must be chosen, and this process considers measurement of input current. As mentioned, mentioned, when the input current is lower than the threshold value (which is 3 A) due to low when the input current is lower than the threshold value (which is 3 A) due to low irradiance level, irradiance and load 1 will be selected. when the input current is highervalue, than FLC 1 and level, load 1FLC will1be selected. However, when However, the input current is higher than threshold threshold load 2 will be selected. FLC 2 andvalue, load 2FLC will 2beand selected.

Energies 2016, 9, 604

7 of 17

Energies 2016, 9, 604

7 of 17 Start

Measure V(k) and I(k)

Calcu late P(k)

ΔP = P(k) – P(k-1) ΔV = V(k) – V(k-1)

E(k) = ΔP/ΔV CE(k) = E(k) – E(k-1)

NO

FLC 1 and Load 1 (Boos t)

YES

Input current > 3 A

FLC 2 and Load 2 (Buck)

Figure 6. Flow chart of the controller. Figure 6. Flow chart of the controller.

4. Simulation Results 4. Simulation Results The PV-basedSEPIC SEPICsystem systemwas wasfirst first modeled MATLAB/Simulink. In this work, The proposed PV-based modeled in in MATLAB/Simulink. In this work, the 2 (KD210GH2 the selected PV panel is a polycrystalline silicon type that produces 210 W at 1000 W/m selected PV panel is a polycrystalline silicon type that produces 210 W at 1000 W/m (KD210GH-2PU, 2PU, Kyocera, Esslingen, Germany). Theparameters main parameters of thePV selected PV panel have been Kyocera, Esslingen, Germany). The main of the selected panel have been tabulated in tabulated in Table 2. Meanwhile, the main components and parameters of shown SEPIC are shown Table 2. Meanwhile, the main components and parameters of SEPIC are in Table 3. in Table 3. For boost operation, the initial duty cycle is set to 0.6 with output voltage will be 39.9 V, and the Tabledown 2. Parameters Kyocera KD210GH-2PU. output voltage will be stepped to 17.7 of V the in buck operation with duty cycle of 0.4. The lower and higher irradiances are set to 200 W/m2 and 700 W/m2., and from both irradiances, the expected Item Value currents produced by the PV system are 1.2 A and 5.63 A. Maximum power (Pmpp ) 210 W Maximum operating (VmppKD210GH-2PU. ) 26.6 V Table 2. Parameters ofvoltage the Kyocera Maximum operating current (Impp ) 7.9 A Short circuit current (Isc ) 8.58Value A Item Open circuit voltage (V ) 33.2 V W oc Maximum power (Pmpp) 210 Temperature coefficient of open circuit voltage ´0.36%/K Maximum operating voltage (Vmpp) 26.6 V Temperature coefficient of short circuit current 0.06%/K Maximum operating current (I mpp ) 7.9 A Temperature coefficient of maximum power ´0.46%/K ˝ Normal operating 25 8.58 C A Short circuit currentcell (Isc)temperature

Open circuit voltage (Voc) Tableof3.open Maincircuit components of SEPIC. Temperature coefficient voltage Temperature coefficient of short circuit current Item Temperature coefficient of maximum powerValue Normal operating cell temperature Inductors 1 and 2 3 mH

33.2 V −0.36%/K 0.06%/K −0.46%/K 25 °C

Capacitors 1 and 2 1000 µF Input 470 µF Table 3. capacitor Main components of SEPIC. Switching frequency 25 kHz Loads 1 and 2 50 Ω and 2 Ω Item Value

Inductors 1 and 2 3 mH Capacitors 1 and 2 to 0.6 1000 F For boost operation, the initial duty cycle is set withµoutput voltage will be 39.9 V, and the Input 470 µ F with duty cycle of 0.4. The lower output voltage will be stepped down to capacitor 17.7 V in buck operation Switching frequency 25 kHz Loads 1 and 2 50 Ω and 2 Ω

Energies 2016, 9, 604

8 of 17

and higher irradiances are set to 200 W/m2 and 700 W/m2 ., and from both irradiances, the expected Energies 2016, 9, 604 8 of 17 currents produced by the PV system are 1.2 A and 5.63 A. For purpose of comparative evaluation, a single-FLC MPPT has been developed too [21]. For purpose of comparative evaluation, a single-FLC MPPT has been developed too [21]. The The membership functions of the single FLC are shown in Figure 7. The membership functions membership functions of the single FLC are shown in Figure 7. The membership functions must must equally be arranged, as to address bothand buck andfunctions. boost functions. As mentioned, output is equally be arranged, as to address both buck boost As mentioned, its output its is changed changed duty to belater added later the pre-defined duty cycle. by dutyby cycle tocycle be added with thewith pre-defined duty cycle. NB

NM

NS

ZE

PS

PM

-2

-1

0

1

2

PB

Degree of membership

1

0.5

0

-5

-4

-3

3

4

5

Error, E (ΔP/ΔV)

(a)

Degree of membership

NB

NM

NS

ZE

PS

PM

-2

-1

0

1

2

PB

1

0.5

0

-5

-4

-3

3

4

5

Change of Error, CE (ΔE)

(b) NB

NM

NS

-2

-1

ZE

PS

PM

0

1

2

PB

Degree of membership

1

0.5

0

-5

-4

-3

Output, ΔD

3

4

5

(c) Figure 7. Membership functions for single-FLC MPPT: (a) error; (b) change of error; and (c) output. Figure 7. Membership functions for single-FLC MPPT: (a) error; (b) change of error; and (c) output.

To further consolidate the contribution of this paper, Dual FLC has been compared with another further consolidate ofof this paper, Dual FLC hasSingle been compared another twoTomore single FLCs withthe thecontribution same number membership functions, FLC 1 and with Single FLC two more single FLCs with the same number of membership functions, Single FLC 1 and Single FLC 2 2 MPPT that have different membership function patterns. Single FLC 1 has membership functions MPPT that have different membership function patterns. Single FLC 1 has membership functions patterned similar to FLC 1 as in Figure 4, while Single FLC 2 has membership functions similar to patterned similar to FLC 1 as in Figure while FLC Single FLC Single 2 has membership functions similar8 to FLC FLC 2 as in Figure 5. The design of4,Single 1 and FLC 2 is shown in Figures and 92 as respectively. in Figure 5. The design of Single FLC 1 and Single FLC 2 is shown in Figures 8 and 9 respectively.

Energies2016, 2016,9,9,604 604 Energies

99ofof17 17

Degree of membership

Energies 2016, 9, 604

9 of 17

NB

Degree of membership

1

NM

NS ZE

PS

NM

NS ZE

PS

NB

1

PM

PB

PM

PB

0.5 0.5

0

-2.00

-1.5

-2.0

-1

-1.5

-1

-0.5

0

-0.5

Error, E0(ΔP/ΔV) Error, E (ΔP/ΔV) (a)

0.5

1

0.5

1.5

1

2

1.5

2

NB

1

Degree of membership

Degree of membership

(a)

NM

NB

1

NM

NS ZE

PS

NS ZE

PM

PS

PM

PB

PB

0.50.5

0 0

-2.0-2.0

-1.5 -1.5

-1-1

-0.5 -0.5

00

0.5 0.5

11

1.51.5

2 2

Change of of Error, Error,CE CE(ΔE) (ΔE)

Degree of membership

Degree of membership

(b) (b) NB NB

1 1

NM NM

NS NS

ZE ZE

PS PS

PM PM

PBPB

0.5

0.5

0

0

0.2

0.3

0.2

0.3

0.4

0.5

0.4

0.5

0.6

0.7

0.6

0.8

0.7

Output, ΔD

0.9

0.8

1

0.9

1

Output, ΔD (c)

(c)

Figure 8. Fuzzy membership functions of inputs and output for single FLC 1: (a) error; (b) change of

Degree of membership

Degree of membership

Figure 8. Fuzzy membership functions of inputs and output for single FLC 1: (a) error; (b) change of Figure Fuzzy membership error;8.and (c) output of singlefunctions FLC 1. of inputs and output for single FLC 1: (a) error; (b) change of error; error;and and(c) (c)output outputofofsingle singleFLC FLC1.1. 1

1

NB

NM

NB

NM

NS

ZE

PS

NS

ZE

PS

PM

PB

PM

PB

0.5

0.5

0

-2.0

-1.5

-1

-2.0

-1.5

-1

0

-0.5

0

-0.5

Error, E (ΔP/ΔV) 0

Error,(a)E (ΔP/ΔV)

(a) Figure 9. Cont.

0.5

0.5

1

1.5

1

1.5

2

2

Energies 2016, 9, 604 Energies Energies2016, 2016,9,9,604 604 NB NB

11

membership of membership Degree of Degree

10 of 17 1010ofof1717 NM NM

NS NS

ZE ZE

PS PM PM PS

PB PB

0.5 0.5

00

-2.0 -2.0

-1.5 -1.5

-1 -1

0 0

-0.5

0.5 0.5

11

1.5 1.5

22

Change of Error, Error, CE CE (ΔE) (ΔE)

(b)

Degree membership ofmembership Degreeof

NB NB

NM NM

NS

ZE ZE

PS PS

PM PM

PB PB

11

0.5 0.5

00

00

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4

Output, Output, ΔD ΔD

0.5 0.5

0.6 0.6

0.7 0.7

0.8 0.8

(c) (c) Figure 9. Fuzzy membership functions of inputs and output for single FLC 2: (a) error; (b) change of Figure 9. Fuzzy membership functions of inputs and output for single FLC 2: (a) error; (b) change of error; and (c) output of single FLC 2. output of of single single FLC FLC 2. 2. error; and (c) output

Input Power(W) Input Power(W)

The simulation work initially considers measurement and analysis at PV input due to effect of The simulation simulation work work initially considers considers measurement and at PV input due to effect effect of The measurement and analysis analysis PV both inputloads due to changed irradiance levels.initially In addition, performance of output voltagesatfor has beenof changed irradiance irradiancelevels. levels. In addition, performance of output voltages for both has been changed In addition, performance of output voltages for both hasloads been analyzed analyzed by undergoing switching operation by load changing circuit. The loads waveforms have been analyzed by undergoing switching operation by load changing circuit. The waveforms have been by undergoing operation load changing The waveforms been captured, captured, andswitching performance of thebydual-FLC MPPTcircuit. is compared with thehave single-FLC MPPT, and by captured, and performance of theisdual-FLC MPPT is single-FLC compared MPPT, with the single-FLC parameters MPPT, by performance of the dual-FLC MPPT compared with the by considering considering parameters such as response time and power loss, both during transient part. Figure 10 considering parameters such as response time andtransient power loss, both during transient part. Figure 10 such as response time power loss, Figure 10 shows waveforms of both input shows waveforms of and input powers forboth bothduring buck and boost part. operations. During buck operation, shows waveforms of input powers for both buck and boost operations. During buck operation, both powers both buck and boost During buck operation, bothatMPPTs toduring control MPPTsfor manage to control SEPICoperations. to achieve maximum power as expected 155 W. manage However, MPPTs manage to control SEPIC to achieve maximum power as expected at 155 W. However, during SEPIC to achieve maximum expectedmuch at 155better W. However, duringtime transient the transient part, the proposedpower MPPTasperforms with response only part, 0.1 ms toproposed achieve transient part, the proposed MPPT performs much better with response time only 0.1 ms to achieve MPPT bettertowith time only 0.1 ms0.4 to achieve steady state lower as compared the steadyperforms state as much compared the response single-FLC MPPT with ms. Consequently, energytoloss steady stateMPPT as compared to the single-FLC lower MPPTenergy with 0.4 ms. Consequently, lower energy loss single-FLC with 0.4 ms. Consequently, loss obtained by the dual-FLC MPPT with obtained by the dual-FLC MPPT with only 2.5 µ J as compared to the single-FLC MPPT with 4.2 µ J. obtained dual-FLC MPPT with only 2.5 µ J as compared to the single-FLC MPPT with 4.2both µ J. only 2.5 µJby asthe compared to the single-FLC µJ. Interestingly, difference between Interestingly, big difference between bothMPPT MPPTswith can 4.2 be seen during boostbig operation. While the dualInterestingly, big difference between both MPPTs can be seen during boost operation. While the dualMPPTs can be seen during boost While the(energy dual-FLC 12 ms to achieve FLC MPPT only needs 12 ms to operation. achieve steady state lossMPPT of 4.8only mJ),needs the single-FLC MPPT FLC MPPT only needs 12 ms to achieve steady state (energy loss of 4.8 mJ), the single-FLC MPPT steady state to (energy of 4.8 single-FLC MPPT contributes to bigger energy loss (up to contributes biggerloss energy lossmJ), (upthe to 210 mJ) due to longer response time (100 ms). contributes loss (up(100 to 210 mJ) due to longer response time (100 ms). 210 mJ) due to to bigger longer energy response time ms). 150 150 100 100 50 50 1.3 1.3

DUAL FLC FLC DUAL FLC

1.35 1.35

1.4 1.4

1.45 1.45

1.5 1.5

1.55

Time 1.55(s) Time (a) (s) (a)

Figure 10. Cont.

1.6 1.6

1.65 1.65

1.7 1.7

FLC 1.75 1.75

1.8 1.8

Input Power(W)

Energies 2016, 9, 604 Energies 2016, 9, 604

11 of 17 11 of 17

170 160 150 140 130

DUAL FLC FLC

120 1.4995

1.5

1.5005

Time (s)

1.501

Input Power (W)

(b) 150

DUAL FLC FLC

100

50

0 0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

Time (s)

Input Power (W)

(c) 42 40 38 36 34

DUAL FLC FLC

1

1.02

1.04

1.06

1.08

1.1

1.12

Time (s) (d) Figure Figure 10. 10. Input Input powers powers under under simulation simulation obtained obtained (a) (a) from from low low to to high high irradiance irradiance or orbuck buckoperation; operation; (b) for the focused transient part in buck operation; (c) from high to low irradiance or boost (b) for the focused transient part in buck operation; (c) from high to low irradiance or boostoperation; operation; and boost operation. operation. and (d) (d) for for the the focused focused transient transient part part in in boost

Figure 11 shows output voltages at both loads 1 and 2 during buck and boost operations, Figure 11 shows output voltages at both loads 1 and 2 during buck and boost operations, respectively. During buck operation with high irradiance, there will be voltage overshoot with respectively. During buck operation with high irradiance, there will be voltage overshoot with smaller value (0.1 V) achieved by the dual-FLC MPPT as compared to the single-FLC MPPT (3.2 V). smaller value (0.1 V) achieved by the dual-FLC MPPT as compared to the single-FLC MPPT (3.2 V). Furthermore, no such oscillation occurs with the dual-FLC MPPT, while the single-FLC MPPT shows Furthermore, no such oscillation occurs with the dual-FLC MPPT, while the single-FLC MPPT shows some oscillations before reaching steady state. The negligible oscillation confirms the effectiveness of some oscillations before reaching steady state. The negligible oscillation confirms the effectiveness using direct duty cycle by the dual-FLC MPPT. During boost operation when irradiance is low, of using direct duty cycle by the dual-FLC MPPT. During boost operation when irradiance is low, negligible undershoot with faster response time (less than 0.04 s) is achieved by the dual-FLC MPPT; negligible undershoot with faster response time (less than 0.04 s) is achieved by the dual-FLC MPPT; whereas the single-FLC MPPT causes a longer time for the load 2 to achieve a stable output voltage whereas the single-FLC MPPT causes a longer time for the load 2 to achieve a stable output voltage with a response time of up to 0.17 s, with an undershoot voltage of 4.2 V. Therefore, from all with a response time of up to 0.17 s, with an undershoot voltage of 4.2 V. Therefore, from all simulation simulation results, clearly the dual-FLC MPPT shows better performance, with negligible oscillation, results, clearly the dual-FLC MPPT shows better performance, with negligible oscillation, small small overshoot, fast response time, and negligible undershoot. overshoot, fast response time, and negligible undershoot.

Energies 2016, 9, 604

12 of 17

Energies 2016, 9, 604

12 of 17 12 of 17

42 40

Voltage load 1 (V)

Voltage load 1 (V)

Energies 2016, 9,44604

38 36

34 32 30

44

42 40 38 36

34 32

0.7

0.8

30

45 40 35 30 25 20 15 10 5 0 -5

Voltage load 2 (V)

Voltage load 2 (V)

0.7

0.9 0.8

1

1.1

Time (s)1.1 (a)Time (s)

0.9

1

1.2 1.2

1.3 1.3

(a) 45 40 35 30 25 20 15 10 5 0 -5

1.3 1.3

1.41.4

1.5 1.5

1.6 1.6

Time (s) Time (s)

1.7 1.7

1.8 1.8

(b) (b) Figure 11. Output voltages of simulation work (a) at load 1 during boost operation; and (b) at load 2

Figure 11. Output voltages of simulation work (a) at load 1 during boost operation; and (b) at load 2 during buck operation. during buck operation.

5. Experimental Results

5. Experimental ExperimentalResults Results

A laboratory prototype of the PV-based SEPIC system was developed as shown in Figure 12,

A laboratory SEPIC system developed shown Figure 12, 12, with same parameters to investigate performance of the dual-FLC A Chroma Programmable prototype of the PV-based was MPPT. developed as shown in Figure Power Supply 62100H-600S (600performance V/25 A/15 kW, ATE Inc., Taoyuan, Taiwan)Programmable with Solar with DC same parameters to investigate ofChroma the dual-FLC MPPT. A Chroma Array Simulation is used as solar simulator. ThekW, dual-FLC MPPT as the proposed algorithm, and alsoSolar DC Power Supply 62100H-600S (600 V/25 A/15 kW,Chroma Chroma ATE Inc., Taoyuan, Taiwan) with V/25 A/15 ATE Inc., Taoyuan, Taiwan) the single-FLC MPPT for comparison purpose, are implemented in a TMS320F28335 eZdsp board Array Simulation Simulation isisused usedasassolar solarsimulator. simulator.The The dual-FLC MPPT as the proposed algorithm, and dual-FLC MPPT as the proposed algorithm, and also (Texas Instruments, Dallas, TX, USA). The same waveforms and parameters as defined in the also the single-FLC MPPT for comparison purpose, are implemented in a TMS320F28335 the single-FLC MPPT for comparison purpose, are implemented in a TMS320F28335 eZdsp eZdsp board simulation work are used. The measured and calculated parameters obtained for all simulation board Instruments, Dallas, TX, USA). same waveforms and parametersasasdefined defined in in the (Texas(Texas Instruments, Dallas, TX, USA). TheThe same waveforms and parameters results and experimental results are summarized in Table 4. simulation work are used. The measured and calculated parameters obtained for all simulation results simulation work are used. The measured and calculated parameters obtained for all simulation and experimental results are summarized in Tablein 4.Table 4. results and experimental results are summarized

SEPIC CONVERTER LA25NP (current sensor )

Load 1

SEPIC CONVERTER LA25NP (current sensor )

Load 1

Load 2

LV25NP (voltage sensor )

IGBT driver circuit

DSP TMS320F28335

Load changing controller

Figure 12. Laboratory prototype of the PV-based SEPIC system. Load 2 LV25NP (voltage sensor )

IGBT driver circuit

DSP TMS320F28335

Load changing controller

Figure SEPIC system. system. Figure 12. 12. Laboratory Laboratory prototype prototype of of the the PV-based PV-based SEPIC

Energies 2016, 9, 604

13 of 17

Table 4. Summary of simulation and experimental results. Energies 2016, 9, 604

MPPT

13 of 17

Simulation

Parameter

Position

Experiment

Table 4. Summary of simulation and experimental Buck results. Boost

Dual-FLC 0.1Simulation ms 12 ms MPPT Position Parameter Response Time (s) Single-FLC 0.4 ms Boost 100 ms Buck PV Input Dual-FLC 0.1 12 Dual-FLC 2.5 ms µJ 4.8ms mJ Response Time Energy Loss (J) (s) Single-FLC 0.4 ms Single-FLC 420 µJ 100 210ms mJ PV Input Dual-FLC 2.5 µ J 4.8 mJ Dual-FLC 0.015 s 0.04 s EnergyTime Loss (s) (J) Response Single-FLC 420 mJs Single-FLC 0.08µsJ 210 0.17 Dual-FLC Load 0.015 s 0.04 s Response Time (s) Dual-FLC 0.10 V ´0.1 V Overshoot/Undershoot Voltage (V) Single-FLC 0.08 s 0.17 s Single-FLC 3.2 V ´4.2 V Load Dual-FLC 0.10 V −0.1 V Overshoot/Undershoot Voltage (V) Single-FLC 3.2 V −4.2 V

Buck

Boost

Experiment 0.01 s 0.02 s 0.2 s Boost 0.26 s Buck 0.01 s 0.02 s mJ 4 mJ 19 0.2 s mJ 860s mJ 0.26293 4 mJ 19 mJ 0.03 s 0.01 s 8600.12 mJ s 293 mJ 0.3 s 0.03 s 0.01 s 0.01 V ´0.35 V 0.12 s 0.3 s 5.3 V ´6.8 V 0.01 V −0.35 V 5.3 V −6.8 V

Figure 13 shows waveforms of input powers for both buck and boost operations. During buck 13 the shows waveformswork, of input for both buckperforms and boostbetter operations. buck time operation,Figure like in simulation thepowers dual-FLC MPPT with During a response operation, like in the work, the dual-FLC better with a response time of by of 0.01 s as compared tosimulation the single-FLC MPPT with MPPT 0.2 s. performs Reduction of energy loss is obtained 0.01 s as compared to the single-FLC MPPT with 0.2 s. Reduction of energy loss is obtained by the the dual-FLC MPPT (only 4 mJ) as compared to the single-FLC MPPT with 860 mJ. During boost, dual-FLC MPPT (only 4 mJ) as compared to the single-FLC MPPT with 860 mJ. During boost, a a significant impact of implementing the dual-FLC MPPT is shown, with only 0.02 s needed to achieve significant impact of implementing the dual-FLC MPPT is shown, with only 0.02 s needed to achieve steady state (lower energy loss of 19 mJ), while the single-FLC MPPT causes a bigger energy loss (up to steady state (lower energy loss of 19 mJ), while the single-FLC MPPT causes a bigger energy loss 293 mJ) to mJ) the due longer response time (0.26 s).(0.26 s). (updue to 293 to the longer response time

(a) Input Power (W)

200 Dual FLC FLC

150 100 50 0 8

9

10

11

12

13

14

Time (s) Input Power (W)

(b) Dual FLC FLC

80 60 40 20 0

9

9.5

10

Time (s)

10.5

11

11.5

(c) Figure 13. Input powers under experimental work obtained (a) from low to high irradiance or buck

Figure 13. Input powers under experimental work obtained (a) from low to high irradiance or buck operation; (b) from high to low irradiance or boost operation; and (c) for the focused transient part in operation; (b) from high to low irradiance or boost operation; and (c) for the focused transient part in boost operation. boost operation.

Energies 2016, 9, 604 Energies 2016, 9, 604

14 of 17 14 of 17

Voltage load 1 (V)

Figure shows output Figure 14 14 shows output voltages voltages at at both both loads loads 11 and and 22 during during buck buck and and boost boost operations, operations, respectively. voltage with thethe dual-FLC MPPT as respectively. During Duringbuck buckoperation, operation,there therewill willbebenonoovershoot overshoot voltage with dual-FLC MPPT compared to the single-FLC MPPT (5.3 V). During boost operation, lower undershoot with faster as compared to the single-FLC MPPT (5.3 V). During boost operation, lower undershoot with faster response time(less (lessthan than0.01 0.01s)s) achieved dual-FLC MPPT; however, the single-FLC response time is is achieved by by thethe dual-FLC MPPT; however, the single-FLC MPPTMPPT takes takes (a response to load 0.3 s)2 for load 2a stable to obtain a stable output voltage. All longerlonger (a response time of time up toof 0.3up s) for to obtain output voltage. All experimental experimental results confirm the of effectiveness of MPPT the dual-FLC MPPT in SEPIC operation to step results confirm the effectiveness the dual-FLC in SEPIC operation to step down and up down and up voltage accordingly. voltage accordingly.

45 40 35 30

Dual FLC FLC

25 9.5

10

10.5

Time (s)

11

11.5

12

(a) Voltage Load 2 (V)

22

Dual FLC FLC

20 18 16 4.8

5

5.2

5.4

5.6

Time (s) (b)

5.8

6

6.2

6.4

Figure 14. Output voltages of experimental work (a) at load 1 during boost operation; and (b) at load Figure 14. Output voltages of experimental work (a) at load 1 during boost operation; and (b) at load 2 2 during buck operation. during buck operation.

Changing the single-FLC MPPT to single-FLC 1 MPPT and single-FLC 2 MPPT, the performance thebeen single-FLC MPPT single-FLC 1 MPPT and single-FLC 2 MPPT, performance of theChanging MPPTs has evaluated andtothis case study is rather important to justify thethe selection of the of the MPPTs has been evaluated and this case study is rather important to justify the selection the membership function pattern when applying FLC MPPT. Figure 15 shows the output voltages atofboth membership function pattern when applying FLC MPPT. Figure 15 shows the output voltages at both loads 1 and 2 during buck and boost operations by comparing the performance of single-FLC 1 with loadsFLC 1 and 2 during buckboost and operation, boost operations by comparing the single-FLC performance of single-FLC 1 with dual MPPT. During the performance of both 1 MPPT and dual-FLC dual FLC MPPT. During boost operation, the performance of both single-FLC 1 MPPT and dual-FLC MPPT are the same due to their similar membership function patterns. During buck operation, there MPPT same due to their similar membership function patterns.toDuring buck operation, will beare no the overshoot voltage with the dual-FLC MPPT as compared the single-FLC 1 MPPTthere (5.3 will be no overshoot voltage with the dual-FLC MPPT as compared to the single-FLC 1 MPPT (5.3 V). V). By using single-FLC 1, there is no undershoot voltage during boost operation but overshoot By usingstill single-FLC 1, there is no undershoot voltage during boost operation but overshoot voltage voltage exists during buck operation. still exists buckoutput operation. Figureduring 16 shows voltages at both loads 1 and 2 during buck and boost operations by Figure 16 shows output at both loads 1 and 2 MPPT. during During buck and boost operations comparing the performance ofvoltages single-FLC 2 with dual-FLC buck operation, the by comparing the performance of single-FLC 2 with dual-FLC MPPT. During buck operation, performance of both single-FLC 2 MPPT and dual-FLC MPPT are the same due to their similar the performance of both single-FLC 2 MPPT and dual-FLC same due to their similar membership function patterns. During boost operation, thereMPPT will beare no the undershoot voltage obtained membership function patterns. During boost operation, there will be no undershoot voltage obtained by the dual-FLC MPPT as compared to the single-FLC 2 MPPT. Single-FLC 2 MPPT also presents a by the dual-FLC MPPT compared the single-FLC MPPT also presents greater time response of as about 0.2 s. Bytousing single-FLC22,MPPT. there isSingle-FLC undershoot2 voltage during boost a greater time response of about 0.2 s. By using single-FLC 2, there is undershoot voltage operation but no overshoot voltage exists during buck operation. Table 5 presents simplified during results boostshows operation no overshoot voltagevoltage exists during buck operation. Table 5 presents simplified that the but occurrence of overshoot and undershoot voltage during buck and boost results that shows the occurrence of overshoot voltage and undershoot voltage during buck and condition for each MPPT algorithm and it clearly shows the need of dual FLC MPPT to buckboost and condition for each MPPT algorithm and it clearly shows the the needinitial of dual FLC MPPT to buck and boost boost output voltage without having energy losses during period. output voltage without having energy losses during the initial period.

Energies 2016, 9, 604 Energies 2016, 9, 604 Energies 2016, 9, 604

15 of 17 15 of 17 15 of 17

V Voltage o lta g e loload a d 1 1(V(V) )

50 50 45 45 40 40 35 35 30 30

Dual FLC Dual FLC Single FLC 1 Single FLC 1

25 25 20 20 9.5 9.5

10 10

10.5 10.5

11 11

Time (s) Time (s)

11.5 11.5

12 12

(a) (a)

V Voltage o lta g e L oLoad a d 2 2(V(V) )

22 22

Dual FLC Dual FLC Single FLC 1 Single FLC 1

20 20 18 18 16 16

4.8 4.8

5 5

5.2 5.2

5.4 5.4

5.6

5.8 5.8

5.6(s) Time Time (s)

6 6

6.2 6.2

6.4 6.4

(b) (b)

Figure 15. 15. Output voltages of of experimentalwork work (comparingDual Dual FLCwith with SingleFLC FLC 1), (a) at load1 Figure Output voltages Figure 15. Output voltages of experimental experimental work(comparing (comparing DualFLC FLC withSingle Single FLC1),1),(a)(a)atatload load 1 during boost operation; and (b) at load 2 during buck operation. during boost operation; and (b)(b) at at load 2 during buck operation. 1 during boost operation; and load 2 during buck operation.

Voltage load 1 (V)

Voltage load 1 (V)

45

45 40

40

35

35 30

30

Dual FLC Single FLC 2

25 9.5

10

25 9.5

10.5

10

11

Time (s) (a)

10.5

11.5

11

Dual FLC Single12 FLC 2

11.5

12

Time (s) Voltage V o ltage lo ad load 2 (V )2 (V)

Dual FLC Single FLC 2 Dual FLC Single FLC 2

(a)

19 18.5 19 18 18.5

17.5 18 17 17.5

16.5 17 16 16.5

15.5 16 4.8

5

5.2

5.4

5.6

5.8

6

6.2

5.4

Time (s) 5.6 (b) Time (s)

5.8

6

6.2

15.5 4.8

5

5.2

Figure 16. Output voltages of experimental work (b) (comparing Dual FLC with Single FLC 2), (a) at load 1 during boost operation; and (b) at load 2 during buck operation. Figure 16. 16. Output voltages of of experimental experimentalwork work(comparing (comparingDual DualFLC FLCwith withSingle SingleFLC FLC2),2),(a)(a)atatload load1 Figure Output voltages 1 during boost operation; and load 2 during buck operation. during boost operation; and (b)(b) at at load 2 during buck operation.

Energies 2016, 9, 604

16 of 17

Table 5. Occurrence of overshoot and undershoot voltage during buck and boost with MPPT algorithms. MPPT Algorithms

Overshoot Voltage Occur during Buck Operation

Undershoot Voltage Occur during Boost Operation

Dual FLC (FLC 1 and FLC 2) Single FLC Single FLC 1 Single FLC 2

No Yes Yes No

No Yes No Yes

6. Conclusions This paper has successfully presented significant work on dual-FLC MPPT for SEPIC. The results obtained, from both simulation and experimental work, clearly show the effectiveness of the dual-FLC MPPT as compared to the single-FLC MPPT. In the dual-FLC MPPT, both FLC parts, assigned for buck and boost operations, significantly overcome the problems of overshoot and undershoot voltages. Overcoming these problems consequently contributes to a reduction of energy losses. In addition, besides addressing MPPT for obtaining the maximum power in steady state operation, the proposed MPPT has successfully proven to function in situations of fast changing irradiance. Acknowledgments: This work is supported by the Fundamental Research Gant Scheme under the Ministry of Higher Education, Malaysia (03-01-14-1413FR). Author Contributions: Tanaselan Ramalu designed and developed the main parts of the research work, including simulation model, experimental set up, and analyses of the obtained results. Tanaselan Ramalu was also mainly responsible for preparing the paper. Mohd Amran Mohd Radzi contributed in simulation, experimental, and writing parts. Muhammad Ammirrul Atiqi Mohd Zainuri, Noor Izzri Abdul Wahab and Ribhan Zafira Abdul Rahman also involved in verifying the work and actively contributed to finalize the manuscript. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4. 5.

6.

7.

8.

9. 10.

Parida, B.; Iniya, S.; Goic, R. A review of solar photovoltaic technologies. Renew. Sustain. Energy Rev. 2011, 15, 1625–1636. [CrossRef] Ding, K.; Bian, X.G.; Liu, H.H.; Peng, T. A MATLAB-simulink-based PV module model and its application under conditions of nonuniform irradiance. IEEE Trans. Energy Convers. 2012, 27, 864–872. [CrossRef] Krismadinataa; Rahim, N.A.; Pinga, H.W.; Selvaraja, J. Photovoltaic module modeling using simulink/matlab. Procedia Environ. Sci. 2013, 17, 537–546. [CrossRef] Qi, C.; Ming, Z. Photovoltaic module simulink model for a stand-alone PV system. Phys. Procedia 2012, 24, 94–100. [CrossRef] Durán, E.; Cardona, M.S.D.; Galán, J.; Andújar, J.M. Comparative Analysis of Buck-Boost Converters Used to Obtain I-V Characteristic Curves of Photovoltaic Modules. In Proceedings of the Power Electronics Specialists Conference, Rhodes, Greece, 15–19 June 2008. Mohd Zainuri, M.; Radzi, M.A.M.; Soh, A.C.; Rahim, N.A. Development of Adaptive Perturb and Observe-Fuzzy Control Maximum Power Point Tracking for Photovoltaic Boost DC-DC Converter. IET Renew. Power Gener. 2014, 8, 183–194. [CrossRef] Taghvaee, M.H.; Radzi, M.A.M.; Moosavain, S.M.; Hizam, H.; Marhaban, M.H. A current and future study on non-isolated DC–DC converters for photovoltaic applications. Renew. Sustain. Energy Rev. 2013, 17, 216–227. [CrossRef] El Khateba, A.H.; Rahim, N.A.; Selvaraja, J. Fuzzy Logic Control Approach of a Maximum Power Point Employing SEPIC Converter for Standalone Photovoltaic System. In Proceedings of the 3rd International Conference on Sustainable Future for Human Security, Kyoto, Japan, 3–5 November 2012. Khateb, A.E.; Rahim, N.A.; Selvaraj, J.; Uddin, M.N. Fuzzy-Logic-Controller-Based SEPIC Converter for Maximum Power Point Tracking. IEEE Trans. Ind. Appl. 2014, 50, 2349–2358. [CrossRef] Salas, V.; Olías, E.; Barrado, A.; Lázaro, A. Review of the Maximum Power Point Tracking Algorithms for Stand-Alone Photovoltaic Systems. Sol. Energy Mater. Sol. Cells 2006, 90, 1555–1578. [CrossRef]

Energies 2016, 9, 604

11. 12. 13. 14. 15. 16.

17. 18. 19. 20. 21. 22.

23.

17 of 17

Liu, C.-L.; Chen, J.-H.; Liu, Y.-H.; Yang, Z.-Z. An Asymmetrical Fuzzy-Logic-Control-Based MPPT Algorithm for Photovoltaic Systems. Energies 2014, 7, 2177–2193. [CrossRef] Cheng, P.-C.; Peng, B.-R.; Liu, Y.-H.; Cheng, Y.-S.; Huang, J.-W. Optimization of a Fuzzy-Logic-Control-Based MPPT Algorithm Using the Particle Swarm Optimization Technique. Energies 2015, 8, 5338–5360. [CrossRef] Lalounia, S.; Rekioua, D.; Rekioua, T.; Matagne, E. Fuzzy Logic Control of Stand-Alone Photovoltaic System with Battery Storage. J. Power Sources 2009, 193, 899–907. [CrossRef] Salam, Z.; Ahmed, J.; Merugu, B.S. The application of soft computing methods for MPPT of PV system: A technological and status review. Appl. Energy 2013, 107, 135–148. [CrossRef] Nabulsi, A.A.; Dhaouadi, R. Efficiency Optimization of a DSP-Based Standalone PV System Using Fuzzy Logic and Dual-MPPT Control. IEEE Trans. Ind. Inform. 2012, 8, 573–584. [CrossRef] Cakmak, R.; Altas, I.H.; Sharaf, A.M. Modeling of FLC-Incremental Based MPPT Using DC-DC Boost Converter for Standalone PV System. In Proceedings of the 2012 International Symposium Innovations in Intelligent Systems and Applications (INISTA), Trabzon, Turkey, 2–4 July 2012. Altasa, I.H.; Sharafb, A.M. A Novel Maximum Power Fuzzy Logic Controller for Photovoltaic Solar Energy Systems. Renew. Energy 2008, 33, 388–399. [CrossRef] Karthika, S.; Rathika, P.; Devaraj, D. Fuzzy logic based maximum power point tracking designed for 10 kW solar photovoltaic system. Int. J. Comput. Sci. Manag. Res. 2013, 2, 1421–1427. Rahmani, R.; Seyedmahmoudian, M.M.; Mekhilef, S.; Yusof, R. Implementation of Fuzzy Logic Maximum Power Point Tracking Controller for Photovoltaic System. Am. J. Appl. Sci. 2013, 10, 209–218. [CrossRef] Salah, C.B.; Ouali, M. Comparison of fuzzy logic and neural network in maximum power point tracker for PV systems. Electr. Power Syst. Res. 2011, 81, 43–50. [CrossRef] Chiang, S.J.; Shieh, H.J.; Chen, M.C. Modeling and Control of PV Charger System with SEPIC Converter. IEEE Trans. Ind. Electron. 2009, 56, 4344–4353. [CrossRef] Hajighorbani, S.; Radzi, M.A.M.; Kadir, M.Z.A.A.; Shafie, S. Dual Search Maximum Power Point (DSMPP) Algorithm Based on Mathematical Analysis under Shaded Conditions. Energies 2015, 8, 12116–12146. [CrossRef] Bo, F.; Lu, Y.; Qiong, C.; Xinchen, C.; Xin, X. A Photovoltaic MPPT Strategy by Modified Fuzzy Control. Appl. Mech. Mater. 2012, 128–129, 15–19. © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).