Three-Phase Grid-Connected PV System Operating with Feed-Forward Control Loop and Active PowerLine Conditioning Using NPC Inverter Leonardo B. G. Campanhol1, Sergio A. Oliveira da Silva2, Vinícius D. Bacon2, Azauri A. O. Junior3 Department of Electrical Engineering Federal University of Technology – UTFPR-AP, Apucarana – PR, Brazil 1 Federal University of Technology – UTFPR-CP, Cornélio Procópio – PR, Brazil 2 University of São Paulo – USP, São Carlos – SP, Brazil 3 e-mails: [email protected]
, [email protected]
, [email protected]
, [email protected]
Abstract— This paper presents a single-stage three-phase gridconnected photovoltaic (PV) system, which is implemented using the three-level neutral-point-clamped inverter. The PV system operates with an additional feed-forward control loop (FFCL), in order to improve the dynamic behavior of both dc-bus voltage and inverter currents, when the PV array is subjected to abrupt insolation changes. In other words, sudden solar radiation changes can cause large voltage oscillations on the inverter dc-bus, interfering in an adequate PV system operation, due to the generation of the inverter current references are computed taking into account the dc-bus voltage control loop. Thus, the use of the FFCL reduces the settling time and overshoot/undershoot of the dc-bus voltage. Moreover, the dynamic response of the currents injected into the grid are also improved. Simultaneously, the PV system is able to perform the active power-line conditioning by suppressing load harmonic currents and performing reactive power compensation. Although the FFCL can be used in conjunction with any maximum power point tracking technique, perturb and observe technique is adopted to generate the dc-bus voltage reference. Simulation results are presented in order the evaluate the performance, as well the effectiveness of the PV system operating with the proposed FFCL. Keywords—PV system, feed-forward control loop, active powerline conditioning, three-level NPC inverter.
In recent years, the generation of electric energy based on the use of renewable energy sources (RES) has increased significantly. This has happened due to growing demand for electricity, as well as to reduce the environmental and economic impacts introduced by power generation based on conventional pollutant energy sources. Thus, there is an increasing trend of change related to the global power generation scenario, in which the use of RES based on hydro, biomass, wind and solar has been highlighted. Regarding the use of solar energy, this RES has been employed through the conversion of photovoltaic (PV) energy into electrical energy. By producing reduced environmental impacts, the generation of this form of energy can be considered clean, due to the reduced environmental impacts involved. In addition, the waste is produced only in the manufacturing
978-1-5090-3474-1/16/$31.00 ©2016 IEEE
process or in the disposal. Thus, the production of electricity by means of PV systems is considered promising among the various existing RES. In PV systems, one or more power conversion stages are required, so that the produced electrical energy can be injected into the utility grid. When the power conversion is performed by means of two stages, usually the first stage is used to step-up the PV array voltage, and the second connects the PV system into the grid -. Considering double-stage PV systems, the dc/dc converter performs the maximum power point tracking (MPPT) control, while the dc/ac performs the current control. In many PV system applications, the boost dc/dc stage conversion is discarded -. In this case, the energy produced is sent to the grid by means of only a inverter. As a result, the power losses decrease, while the system efficiency increases . In addition, the control loops employed to perform the MPPT, as well as the inverter current control are carried out in only one power conversion stage. Thereby, the speed of the MPPT control loop (dc-bus control loop) must be much slower than that of the current control loop, in order to ensure undistorted currents injected into the grid . Generally, the PV arrays are subjected to non-uniform insolation, which can interfere in the amount of energy generated by the PV system. On the other hand, abrupt insolation changes can cause large oscillations in the dc-bus voltage. As the dc-bus power balance controls the inverter current amplitudes, the aforementioned oscillations can interfere in a suitable PV system operation. In other words, large voltage variations in the dc-bus can interfere in an adequate computation of the inverter current references. Most of MPPT techniques do not taken into account the effects of abrupt solar radiation on the PV system performance. In , comparative analyses involving a single-stage PV system operating with both perturb and observe (P&O) and particle swarm optimization MPPT techniques were performed. Although the problem associated with partial shading has been taken into account, no procedure was adopted to mitigate the problems associated to fast solar radiance changes.
In , a MPPT technique based on modified incremental conductance has been proposed to improve the stability of the MPPT method, when a single-stage grid-tied PV system is subjected to sudden insolation changes. The modified MPPT method detects the fast decreasing of the PV array output power at the moment that sudden insolation change occurs. However, large dc-bus voltage oscillation can still be observed during the aforementioned transient. Furthermore, the proposed MPPT method is not able to detect fast increasing of the PV array output power, taking into account this effect can also affect the proper operation of the PV system. In this paper the implementation of a feed-forward control loop (FFCL) is proposed, which is used to overcome the problems related to abrupt solar radiation changes in gridconnected PV systems. The FFCL is not tied to any MPPT technique. It acts in conjunction with the inverter dc-bus voltage controller improving the dynamic responses of both dc-bus voltage and inverter currents during atmospheric transient conditions. In addition, reducing the inverter dc-bus voltage oscillations, the risk of over voltage rating in the power semiconductors decreases. Power Supply
Lsa Lsb Lsc
Besides active power injection into the utility grid, in this paper the three-phase single-stage PV system is also able to perform the active conditioning, resulting the suppression of the load harmonic currents and the compensation of the reactive power , , -. The grid-connected PV system is implemented by means of a three-level neutral-point-clamped (NPC) inverter , , where the synchronous reference frame (SRF) algorithm is employed to generate the non-active current components. In addition, the synchronous coordinates, ( ) and ( ), are achieved from the utility voltage phaseangle ( ), which is estimated using a phase-locked loop (PLL) system . This paper is organized as follows: In Section II is presented the PV system description, including the SRF-based current reference generator (CRG), and the dc-bus voltage and current controllers. Section III presents the FFCL and the adopted PLL system. In Section IV the performance and feasibility of the three-phase single-stage grid-tied PV system are evaluated by means of simulation results. Finally, the conclusions are presented in Section V.
vsa i sa vsb i sb vsc i
iLa LLa iLb LLb iLc LLc iLn
isn Control System
ipv vdc1 vdc2 vsa vsb vsc iLa iLb iLc ica icb icc
* MPPT vdc (P&O) -+
P sin(θ ) pll L cos(θ pll) L
Current Reference Generator (SRF-based algorithm)
PI-MR P W controller M
s1a s2a s3a s4a
PI-MR P W controller M
s1b s2b s3b s4b
PI-MR P W controller M
s1c s2c s3c s4c
RLfc RLfb RLfa L-Filters
vpv ipv s1a
Parallel NPC Converter
PV1 PV2 PV3
dc-bus PV array
Fig. 1. Single-stage three-phase PV system scheme.
The grid-tied PV system implemented in this paper is presented in Fig. 1. It can be applied to three-phase four-wire systems. As shown in Fig. 1, the single-stage PV system is composed of a PV array and a three-level NPC inverter is tied to the grid using L-filters. This NPC converter topology has been widely used in transformerless applications involving PV systems , , . The PV array is comprised of twenty solar panels connected in series (SolarWorld SW 245). At MPP, each solar panel provides 30.8V when operating in standard test conditions (STC), such that the total dc-bus voltage is around 616V. Therefore, the PV array can be directly connected to the dc-bus inverter.
A. Current Reference Generator Algorithm A CRG algorithm shown in Fig. 2 is employed to compute the grid-tied inverter current references ( ∗ , ∗ , ∗ ), which includes both the active and non-active current components. The CRG is based on the synchronous reference frame (SRF) method. In the SRF-based algorithm, the load currents are transformed from the -axes into 0-axes. Next, the currents are transformed into the two-phase synchronous reference frame ( -axes). These transformations are given by is (1) and (2), respectively. In the -axes, the current composed of both continuous and alternate components. The ) represents the load fundamental continuous current ( ) represents the active component, and the alternate current ( is load harmonic components in the -axis. The current extracted using a high-pass filter (HPF), where its
implementation makes use of a low-pass filter (LPF), as shown represents the sum of the in Fig. 2. In addition, the current load fundamental reactive component and the harmonic represents the zero-sequence components in the q-axis, while component of the neutral wire. Therefore, in order to guarantee and must be totally compensated unity power factor, both . in conjunction with
represented by the inductances ( , , ) and resistances ( , , ) of the L-filters, in conjunction with the dc-bus voltage ( /2). The PWM inverter static gain, which is represented by , is calculated taking into account the triangular carrier peak . Since alternate components compose the inverter current references ( ∗ , ∗ , ∗ ), the classical proportional-integral (PI) controller does not guarantee null steady-state error. Thus, the PI multi-resonant (PI-MR) controllers are adopted. The transfer function of the PI-MR controller is given by (5). Thus, due to the high open-loop gains at resonant frequency ( ), zero steady-state error is achieved.
Fig. 2. Three-phase current reference generator algorithm.
1 1 − 2 2 √3 √3 − 2 2 1 1
2 0 3 1
√2 √2 = −
The inverse transformations from the rotating reference frame into the reference frame and from the 0 reference frame into the reference frame are given by (3) and (4), respectively. Thus, the final inverter current references ( ∗ , ∗ , ∗ ) are obtained. =
∗ ∗ ∗
√2 1 2 1 √3 − 2 3 2 √2 1 √3 1 − − 2 2 √2
+ ( )+
) ) ( )(
In (5) and represent the respective current PI is the resonant gain at = ; is the controller gains; utility fundamental frequency; and = 1, 3, 5, 7, and 9, represents the resonant terms. The first order component (m = 1) represents the fundamental active and reactive components. C. Voltage Control Loops of the PV System dc-Bus In Fig. 3 two dc-bus voltage PI controllers are presented. The first, denominated PI , is used to regulated the total dc-bus voltage at a constant reference ∗ , which is obtained from the P&O-based MPPT algorithm, as shown in Fig. 1. This control is performed by means of the output signal , which represents the total energy produced by the PV system plus the energy demanded by the PV system to compensate losses of both the filtering inductors and power switches. The quantity is used in the CRG algorithm, as shown in (3). , is used to The second dc-bus controller, denominated PI compensate the dc-bus voltage unbalances. In this case, the − ) is difference between the dc-bus voltages ( measured and compared to zero. The PI output signal is also used in the CRG algorithm, as shown in (4).
and are As can be noted in (3) and (4), the currents extracted from their respective dc-bus voltage control loops is (Fig. 3). Furthermore, the feed-forward current (FFC) obtained from the FFCL. The difference between and ( − ) shown in (2) performs the dc-bus active power balance of the PV system, as well as controlling the active power demanded to regulate the dc-bus voltage, considering the compensation of losses in both the L-filters and power switches. B. Current Control Loop Fig. 3 shows the block diagram of the current control loop used in each phase of the PV system. The physical system is
Fig. 3. Block diagram of the current and dc-bus control loops.
FEED-FORWARD CONTROL LOOP
The maintaining the dc-bus voltage is realized by means of the dc-bus voltage controller in order to ensure PV system power balance. On the other hand, this controller is not fast enough to avoid voltage oscillations when sudden insolation changes occur. This happens because the voltage control loop must be designed to be much slower than the current control loops to
guarantee no interference with each other, which could affect the quality of the current waveform injected into the grid. In other words, the FFCL increases the speed of the dc-bus power balance. As a result, the dynamic response of the dc-bus voltage during fast solar radiation transients is improved. The feed-forward control action is implemented by means of the FFC shown in Fig. 2. Therefore, the FFCL acts through the CRG to attenuate the amplitude oscillations that occurs in the dc-bus voltage. In other words, the FFCL estimates the PV array available energy. Thus, the inverter active current component that will be injected into the grid can also be previously estimated. The PV array ( follows:
) active power can be computed as
Assuming that the inverter currents are balanced and the PV system handles active energy, the active power ( ) injected into the grid can be found as: 3
(7) 2 where and represent the rms values of the fundamental utility voltage and current, respectively, and are their respective peak amplitudes. and =3
Supposing that the PV system operates in ideal conditions, such that the system losses are disregarded. Thus, the power = . produced by the PV panels is sent to the grid, i.e., Therefore, it is possible to estimate the utility peak current ( ) using (8). The voltage and current of the PV system are measured, while is be estimated from the PLL system, as will be discussed in the next subsection. =
acts in the SRF, as shown in The feed-forward current Fig. 2. Thus, must be transformed from the abc-axes into the dq-axes as given by: =
As can be noted, to implement the FFCL, no additional voltage or current transducers are needed, since the parameters that compose have been previously measured or estimated. A. PLL system The there-phase PLL system used in this work, named AFPSD-3pPLL is shown in Fig. 4. A positive-sequence detector (PSD) and non-autonomous adaptive filters (AF) act together with the three-phase power-based PLL system (3pPLL) . The AF gain ( ) shown in Fig. 4 are designed taking into account the stability analysis presented in . The PI gains and , as well as the adopted gain = ⁄ are is the sampling time of the presented in the Table I, where digital signal processor (DSP) and μ is the adaptation step-size
of the AFs. The gain = = + is the positivesequence amplitude of the three-phase utility voltage, such that = ( + + )/3 and = ( + + )/3, where , , and , , are the AF weights in the abc stationary reference frame.
2K ___c w1 s
w 2K ___c 2
s x2 AF + vsa w1a vs1a N x1a AFa w 2a x2a + vsb w1b P vs1b N x1b AFb S w 2b x2b + D vsc w1c vs1cN x1c AFc w2c x2c
ωff ω ,
^ ^ ω 1 θpll sin (θ^ - π/2) x1a __ pll s x2a -sin (θ^pll) x 1b ^ ^
sin (θpll) ^
sin (θpll +2π/3)
sin (θpll -7π/6)
-sin (θ^pll -2π/3) sin (θ^pll + π/6) ^
-sin (θpll +2π/3)
x2b x1c x2c
Fig. 4. AF-PSD-3pPLL system.
Fig. 1 presents the PV system scheme implemented in this paper. MATLAB® and Simulink® software tools were used to implement the computational simulations, as follows: 3L-NPC inverter, PV array model, FFCL, P&O-based MPPT, CRG, PLL system, and current and dc-bus voltage controllers. Furthermore, to obtain the simulation results similar to experimental results, the main characteristics of a real PV system were considered, as follows: conditioning signal boards (anti-aliasing filters), semiconductor device dead-time, DSP analog/digital conversion, and control system discretization. In addition, the PV array was implemented using the equivalent electric circuit presented in . In Table I the PV system parameters and the PI gains used in the computational simulations are summarized. Fig. 5 shows the static behavior of the PV system considering three different operation modes (OPM), which are described as follows: OPM 1: injection of only active power into the grid; OPM 2: injection of active power in conjunction with active filtering; OPM 3: only active power-line conditioning. The following quantities are shown: grid voltages ( , , ), grid currents ( , , ), inverter currents ( , , ), and currents of the load ( , , ). The OPM 1 is shown in Fig. 5 (a), where the total active power generated by the PV array ( = 4900 W) is provided to the grid. Thus, the grid currents are sinusoidal, as well as in opposite phase with respect to the grid voltage. In the OPM 2 (Fig. 5 (b)), nonlinear loads are connected to the grid. As the load active power is lower than , part of is consumed by the load and the remaining power is injected into the grid. Fig. 5 (c) presents the OPM 3, where is zero. Thereby, the PV system carries out only active filtering. As can be seen, the compensated source currents are always sinusoidal and in phase with the utility voltages. The total harmonic distortions (THD) of both grid and load currents are presented in Table II. As can be noticed, the THDs of the compensated source currents were significantly reduced (lower than 5%), in all OPMs. Fig. 6 presents the PV system operating without the FFCL, considering the PV array subjected to abrupt insolation changes.
During this time interval, in which the PV array is disconnected, the PV system operates in OPM 1 and the dc-bus voltage reference is set in ∗ = 460 V. When the PV array is reconnected, the PV system achieves the MPP around of 70 s, such that ∗ = 616 V. In this case, ∗ is defined by the P&Obased MPPT algorithm. As can be observed, during the transients the dynamic of the current is very slow interfering in the dc-bus voltage overshoot/undershoot and setting time.
Up to 10 s, the PV system is operating normally. At this moment the PV array is disconnected and reconnected again after 35 s. Details related to the transients can be observed in Fig. 6 (b) and (c), where oscillations are observed in the dc-bus voltage.
TABLE I – PV SYSTEM PARAMETERS AND CONTROLLER GAINS Nominal utility voltages (rms) Inverter inductive filters Resistance of the inverter inductive filters Total dc-bus capacitor MPP dc-bus voltage NPC inverter switching frequency A/D converters sampling frequency Step size voltage (MPPT-P&O) Sampling time (MPPT-P&O)
= 127.27 V = 1.7 mH = 0.2 Ω = 2350 µF = 616 V = 20 kHz = 60 kHz ∆ = 1V = 166.67 ms =2x10-4
Fig. 7 shows the same tests considering the FFC acting to improve the dynamic response of the PV system. Details related to the transients can be observed in Fig. 7 (b) and (c), where no oscillations are observed in the dc-bus voltage. As can be seen, during the transients, the dynamic of the current is fast enough to reduce strongly the dc-bus voltage setting time and overshoot/undershoot. TABLE II – GRID AND LOAD CURRENTS THDS
PWM gain Nonlinear loads Load Phase “a” Phase “b” Phase “c” R=12.5 Ω R=19.5 Ω R=58 Ω Single-phase L=15.6 mH L=24.2 mH C=940 µF full bridge rectifiers RL and RC loads = 1.3 mH = 3 mH = 6.1 mH PI-MR current controllers 4 crossover frequency = 1.5708e rad/s o phase margin = 85 current controller gains = 439.18 = 6.5476e5 = 15700; = 15627; Current MR controller gains = 15482; = 15265; = 14975 dc-bus voltage controllers voltage controller gains = 0.1797 = 1.3615 crossover frequency = 28.274 rad/s phase margin = 75o unbalance controller gains = 0.0523 = 0.0746 crossover frequency = 8.0784 rad/s phase margin = 80o
Total harmonic distortion (THD %)
This paper presented a single-stage three-phase grid-tied PV system using a three-level NPC inverter. In order to improve dynamic response of the dc-bus voltage and, hence, avoid large oscillations when the PV panels is subjected to solar irradiation changes, a FFCL control loop is proposed. The FFCL contributes to increase the speed of the dc-bus power balance, and acts to generate the inverter current references faster. As a result, the PV system performance was improved. As could be noted, the dc-bus voltage settling time, as well as the overshoot/undershoot were strongly reduced when the FFCL was employed. In addition, the PV system active power-line conditioning capability was verified, resulting in the compensation of load reactive power, as well as suppression of load harmonic currents. As a consequence, an effective power factor correction was obtained.
PLL parameters and controller PLL controller gains = 141.7; = 7777.4 AF gain = 300 step-size adaptation parameter = 0.005
(a) (b) (c) Fig. 5. PV system grid voltages ( , , ), grid currents ( , , ), inverter currents ( , , ) and load currents ( , , ): (a) OPM 1: injected active power ( = 4900W, = 0W); (b) OPM 2: active power injection and active conditioning ( = 4900W); (c) OPM 3: active conditioning ( = 0W).
Ppv Time [s]
(a) (b) (c) and PV array power ( ) for abrupt solar radiation change considering OPM 1 without FFCL Fig. 6. Total dc-bus voltages ( ) and ( ∗ ), current , current (T=25 C): (a) solar radiation transients, (b) detail of the solar radiation sudden decreasing, (c) detail of the solar radiation sudden increasing. 10002 W/m
iff Ppv Time [s]
(a) (b) (c) and PV array power ( ) for abrupt solar radiation change considering OPM 1 with FFCL Fig. 7. Total dc-bus voltages ( ) and ( ∗ ), current , current (T=25 C): (a) solar radiation transients, (b) detail of the solar radiation sudden decreasing, (c) detail of the solar radiation sudden increasing. 
S.B. Kjaer, J.K. Pedersen, and F. Blaabjerg, “A review of single-phase grid-connected inverters for photovoltaic modules,” IEEE Trans. Ind. Appl., vol. 41, pp. 1292-1306, Sept.-Oct. 2005. Y. Kim, H. Cha, B.M. Song, and K.Y. Lee, “Design and control of a gridconnected three-phase 3-level NPC inverter for building integrated photovoltaic systems,” in Proc. IEEE PES Innovative Smart Grid Technologies (ISGT), 2012, pp. 1-7. F.J. Lin, K.C. Lu, T.H. Ke, B.H. Yang and Y.R. Chang, “Reactive power control of three-phase grid-connected PV system during grid faults using Takagi–Sugeno–Kang probabilistic fuzzy neural network control,” IEEE Trans. on Ind. Electron., vol. 62, no. 9, pp. 5516-5528, Sept. 2015. M.M.R. Singaravel, and S.A. Daniel, “MPPT with single DC–DC converter and inverter for grid-connected hybrid wind-driven PMSG–PV system,” IEEE Trans. on Ind. Electron., vol. 62, no. 8, pp. 4849-4857, Aug. 2015. S.A.O. Silva, L.B.G. Campanhol, V.D. Bacon, and L.P. Sampaio, “Single-phase grid-connected photovoltaic system with active power line conditioning,” Brazilian Power Electronics Journal, vol. 20, no. 1, pp. 8– 18, Fev. 2015. F.M. Oliveira, S.A.O. Silva, F.R. Durand, L.P. Sampaio, V.D. Bacon and L.B.G. Campanhol, “Grid-tied photovoltaic system based on PSO MPPT technique with active power line conditioning,” IET Power Electronics, vol. 9, no. 6, pp. 1180–1191, Ap. 2016. M.C. Cavalcanti, A.M. Farias, K.C. Oliveira, F.A.S. Neves, and J.L. Afonso, “Eliminating leakage currents in neutral point clamped inverters for photovoltaic systems,” IEEE Trans. on Ind. Electron., vol. 59, pp. 435443, Oct. 2011. N. Vázquez, M. Rosas, C. Hernández, E. Vázquez and F.J. Perez-Pinal, “A new common-mode transformerless photovoltaic inverter,” IEEE Trans. on Ind. Electron., vol. 62, no. 10, pp. 6381-6391, Oct. 2015.
Powered by TCPDF (www.tcpdf.org)
W. Li, Y. Gu, H. Luo, W. Cui, X. He, and C. Xia, “Topology review and derivation methodology of single-phase transformerless photovoltaic inverters for leakage current suppression,” IEEE Trans. on Ind. Electron., vol. 62, no. 7, pp. 4537-4551, Jul. 2015. Y. Tang, W. Yao, P.C. Loh and F. Blaabjerg, “Highly reliable transformerless photovoltaic inverters with leakage current and pulsating power elimination,” IEEE Trans. on Ind. Electron., vol. 63, no. 2, pp. 1016-1026, Feb. 2016. W. Libo, Z. Zhengming, and L. Jianzheng, “A single-stage three-phase grid-connected photovoltaic system with modified MPPT method and reactive power compensation,” IEEE Trans. on Power Electron., vol. 22, no. 4, pp. 881-886, Dec. 2007. C. Roncero-Clemente, O. Husev, E. Romero-Cadaval, J. Martins, D. Vinnikov, and M. I. Milanes-Montero, “Three-phase three-level neutralpoint-clamped qZ source inverter with active filtering capabilities,” in Proc. Compatibility and Power Electronics, 2015, pp. 216-220. T. Geury, S. Pinto, and J. Gyselinck, “Three-phase power controlled PV current source inverter with incorporated active power filtering,” in Proc. Industrial Electronics Conference (IECON), 2013, pp. 1374-1379. C.H. Chang, Y.H. Lin, Y.M. Chen, and Y.R. Chang, “Simplified reactive power control for single-phase grid-connected photovoltaic inverters,” IEEE Trans. on Ind. Electron., vol. 61, no. 5, pp. 2286-2296, May 2014. S. Sezen, A. Aktas, M. Ucar, and E. Ozdemir, “A three-phase three-level NPC inverter based grid-connected photovoltaic system with active power filtering,” in Proc. Power Electronics and Motion Control Conference and Exposition, 2014, pp. 1331-1335. V.D. Bacon, and S.A.O. Silva, “Performance improvement of a threephase phase-locked-loop algorithm under utility voltage disturbances using non-autonomous adaptive filters,” IET Power Electronics, vol. 8, no. 11, pp. 2237-2250, Nov. 2015. S. Buso, and P. Mattavelli, Digital Control in Power Electronics. Lincoln: Morgan & Claypool Publishers, USA, 2006.