Cuckoo Search Optimization Technique for Multi ...

3 downloads 0 Views 218KB Size Report
Search Algorithm (CSA) while considering the objective load fitness criteria. This .... In this work we divide the 24-hour time period into On-peak and Off-.

Cuckoo Search Optimization Technique for Multi-objective Home Energy Management Adia Khalid1 , Ayesha Zafar1 , Samia Abid1 , Rabiya Khalid1 , Zahoor Ali Khan2 , Umar Qasim3 , and Nadeem Javaid1,∗

Abstract Increasing demand of power and emergence of smart grid has gain maximum attention of researchers which has further opened new opportunities for Home Energy Management System (HEMS). HEMS under Demand Response (DR) helps to reduce the On-peak hour load by shifting the load toward the Off-peak hours. This load shifting strategy effects the user comfort, however in return DR gives them incentives in term of electricity bill reduction. Consumer electricity cost and peak load have a tradeoff, to sort out this situation an efficient system is required. In this paper, we present a multi-objective HEMS to schedule home appliances using Cuckoo Search Algorithm (CSA) while considering the objective load fitness criteria. This proposed load fitness criteria effectively reduces the cost and peak load. Simulations are performed to verify the generic behavior i.e., system performance on any price tariffs. For this purpose, results are validated for three price signals: day-ahead Real Time Peak Price (RTP), Time of Use (TOU) and Critical Peak Price (CPP).

1 Introduction The strategies of Demand Side Management (DSM) planning, development and monitoring focus on power load management i.e., maintaining a balance between consumers demand and its respective supply. Electricity is extensively used in commercial, industrial and residential buildings. Due to the irregularity of electricity consumption pattern, residential area is highly distracted area. This irregularity opens the room for research in Home Energy Management System (HEMS) to monitor and control the home load. One of the key objective of HEMS is to establish bal1 COMSATS

Institute of Information Technology, Islamabad 44000, Pakistan

2 Computer Information Science, Higher Colleges of Technology, Fujairah 4114, United Arab Emi-

rates 3 Cameron Library, University of Alberta, Edmonton, AB, T6G 2J8, Canada ∗Correspondance: www.njavaid.com, [email protected]

1

2

Adia Khalid et al.

ance between demand and supply which is a challenging task especially during the high power demanded hours. It may also be avoided by employing DSM strategies at consumer end and maintaining a normalized consumption pattern. This would help the utility to provide electricity to consumer at all times. It may also be accomplished either through, extra generation or power load shifting; extra generation pay more cost and load shifting discomfort the user. Regardless of this all many techniques have been developed that shift the high peak hours load toward the low peak hours efficiently. Avoiding the peak during the low peak hours is quite challenging. This alteration in power load is key feature of DSM which could be accomplished under six generic categories [1]: valley filling, load shifting, energy efficiency, peak clipping, strategic conversation and strategic load growth. This load alteration is practically possible through Demand Response (DR), which encourage the consumers to play a significant role in HEMS in response of the smart meter pricing infrastructure based on: Time of Use (TOU), Real Time Price (RTP), Critical Peak Rebate (CPR), Critical Peak Price (CPP), and many more. Different optimization techniques are adopted to schedule the load under DR as discussed in [2]-[14], so that electricity consumption pattern could be in a normalized form. These scheduling techniques help in reducing the electricity cost and Peak to Average Ratio (PAR). The proposed architectures in [2]-[14] either focus on cost reduction, PAR or user comfort, all these objectives have tradeoff, which encourage an effectual system. In this regards we proposed a multi-objective HEMS for load management, PAR and cost reduction while sacrificing the user comfort. It is worth mentioning here that load shifting can create peaks during the low peak hours if proper strategy is not applied. In our scheme peak load is avoided by our proposed objective load curve, which used as a constraint of objective cost reduction, help in PAR reduction. Moreover, Cuckoo Search Algorithm CSA) is adopted to schedule the home appliances. Rest of the paper is organized as: Section 2 reflects the state of the art HEMS, and Section 3 incorporates the system model of proposed scheme using CSA. Results are demonstrated on the basis of simulation in Section 4. Finally, conclusion of the research is discussed in Section 5.

2 Related work A considerable amount of research work on HEMS has been published, and integrated in real world. DR program is expected as vital for HEMS; whose purpose is electric appliance scheduling in order to reduce the cost. This scheduling is challenging and need optimal solution [2]. Basics of HEMS do not only focus on cost reduction, also user comfort. These two objective have tradeoff, so in order to accomplish one task other should sacrifices. Besides of all these PAR cannot be ignored, as it has direct impact on user cost. Literature works take HEMS as single-objective or multi-objective.

Cuckoo search optimization technique for multi objective home energy management

3

As a single-objective researchers in [2] target cost minimization. Huang et al., formulated scheduling problem as mixed discrete continuous optimization nonlinear problem, and solved the problem by gradient-based repair PSO optimization technique. Article [3], optimized the energy consumption by considering the interaction between two parties through two-step centralized game. On the basis of this centralized game model, game-theoretic energy schedule model is proposed to reduce the PAR. Furthermore, authors in this article also proposed new price model and objective function. Simulation results show 19% decrease in PAR then unscheduled load and each user pay cost according to the load and overall demand. As discussed above user comfort is also important factor of HEMS, in this perspective a bidirectional framework is proposed by [4], sparse pattern is created through Nash equilibrium using combine effect of dual fast gradient and convex relation to increase the user comfort. This user comfort is achieved by tackling the sub-problems. Moreover, to improve the search efficiency, and to minimize the PAR a newton method is employed. User demanded load is important factor in this regard, Cakmak et al.,, adopted the balanced load curve in article [5] while considering the user preference. The optimum scheduling is performed using CSA, which reduced the 22% peak load through load shifting. Authors in article [6] implement scheduling and also energy storage devices along with introducing the load cluster. In this research paper researchers predict the customer aggregated load on the basis of previous load, and grouped the consumer loads and associated batteries to schedule the appliances using day-ahead price tariff. New binary backtracking search algorithm is proposed by [7] for real time optimal schedule controller. This article shift the home appliance form the On-peak hours using the load limit. Simulation results are contacted and compared with Particle Swarm Optimization (PSO). Proposed scheme has reduced 21.07% electricity price per day and outperformed than PSO. Research article [8], presented HEMS working based on the electricity price forecasting error and system load. These uncertainties in electricity price and load are handled through a chance constrained optimization-based model. Optimum solution is find out through two-point estimate method and improved gradient based PSO. To tackle the multi-objectives with tradeoff a considerable work has been done. In this regard multi-objective genetic algorithm is presented in article [9] to solve the conflicting objectives: 1) objective load curve and electricity bill minimization. Experimental results depicted that multi-objective outperformed rather than if problem salved as single objective. Multi-objective mixed integer linear programming is adopted by [10] to minimize the electricity bill and peak load, while considering the user load demand and preference. Different experimental results for different scenario shows heights 48.6% PAR reduction, where TOU price tariffs was used.

4

Adia Khalid et al.

3 System model The proposed optimization algorithm based on CSA systematic visual representation is given in Fig. 3. The given system model consists of generation unit and consumer unit. Where, our focused area is residential area, further focus is narrowed down towards the single home. As shown in this figure, the considered home consists of smart meter, smart appliances and a home energy management controller. The given energy management controller is embedded with scheduler which first defines the objective load curve based on the user demand and price signals. After that appliances are scheduled according to the demarcated load curve. This proposed comprehensive home load management strategy is used to solve the DSM problem regarding the electricity load demand of On-peak hours. The main focus of this study is on consumer electricity bill reduction through load shifting and PAR reduction. Home energy management controller gives the objective load curve according to the user demanded load and price signals. This strategy is helpful for the users and utility which have day-ahead or seasonal price tariffs. The tactical plaining of home load control unit is mathematically formulated in equation 1, which helps in creating the objective load curve by adapting the strategy defined in article [11]. According to the authors in [11], in order to reduce the electricity bill; chosen objective load curve should be inversely proportional to the electricity market prices. In this work we divide the 24-hour time period into On-peak and Offpeak hours using the price tariffs and load is fulfilled in each hour using equation 1, this equation is formulated here with the modification in the formula given in article [12], further this formula is used as constraint to achieve the objective load curve.

3.1 Cuckoo search for HEMS Meta-heuristic cuckoo search optimization algorithm proposed by [13] is adapted for DSM scheduling. This algorithm is based on some cuckoo species obligate brood parasitic behavior combined with Levy Flight behavior of some fruit flies. CSA optimization's start work with random initialization of nest using the following three rules: 1) each cuckoo lays one egg only once a time and dump into randomly chosen nest; 2) the best nest is selected for next generation; 3) host nests are taken fixed, and probability of finding the cuckoo's egg by host birds considered as 0.4 after performing different experiments. In our home energy management scenario host nests reflects the search space and eggs laid by cuckoo are taken appliance ON and OFF status, if host bird recognized the cuckoo's egg then status will be OFF otherwise ON. Total number of eggs are equal to the total number of appliances. The fitness Fitness criteria for selecting the best nest is formulated as follows: ( ELiεnest ≥ γ × mean(ELunsch ) − std(ELunsch ) Ho f fp Fitness = min (1) iεnest unsch ≥ std(EL ) H on p EL

Cuckoo search optimization technique for multi objective home energy management

5

Fig. 1 System Model diagram

Where ELunsch is the list of unscheduled load, and H o f f p and H on p represent Off and On-peak hours respectively. Current hour is considered as H o f f p i f EPh ≥ mean(EP) else H on p .

3.2 Multi objective optimization of proposed HEMS Above mention system model is multi objective problem with targets: cost minimization and PAR reduction. The multi objective problem is formulated as: minimize

Ftotal = Fcost + FPAR

(2)

The objective function of overall cost minimization for 24 hours is mathematically represented as: 24 M

minimize

Fcost =

∑ ∑ (EPh × AppaPrate × λ )

(3)

h=1 a=1

Where EPh is the electric price for a particular hour, and AppaPrate is the power rate and λ = [0, 1] is ON or OFF status of appliance. Second objective minimization of PAR is formally written as follows:

6

Adia Khalid et al.

minimize

H 2 H FPAR = max(ELsch ) /(sum(ELsch )/24)2

(4)

H is the complete list of per hour load which is calculated using In this equation ELsch equation 5 where ∀H ∈ 1, 2, 3, ..., 24. M

EL =

∑ AppaPrate × λ

(5)

a=1

3.2.1 Model transformation There is a tradeoff between cost and PAR, for simplicity above defined objective function in equation 2 is transformed into single objective using weight function as formulated in article [14]. In this research work we just consider the cost function and ignore PAR. minimize Ftotal = αFcost + (1 − α)FPAR

(6)

where α = 1 is defined as our preference factor.

4 Results and discussion In order to verify the effectiveness of proposed home load management system under DSM, different experiments are carried out for three different price tariffs which includes: RTP, TOU and CPP. Simulations are carried out for a single home which contains 15 appliances as given in Table 1. The considered appliances are further categorized into three types: 1). schedulable interruptible 2). schedulable interruptible and 3). fixed. Appliances classified as schedulable interruptible can be scheduled with interrupt any time during the day. Schedulable un-interruptible appliances (washing machine and cloth dryer) are shift able without interruption during the operational cycle with the restriction that cloth dryer will always start operation when washing machine will complete its desired operational time. Appliances considered as fixed cannot be scheduled and turn ON according to user demand. Results discussed in this section contain cost and PAR of a single home for a particular day using three price tariffs. Figs. 2 and 3 show the pattern of electricity load and cost per hour along the price tariff on the right y-axis. RTP tariff is taken from article [15] depicted in Fig. 2(a), TOU and CPP taken from [16] are highlighted in Fig. 2(b) and 2(c) respectively. Fig. 2 demonstrated that objective load curve is efficaciously achieved using equation 1. Graphical results deliberate that load from On-peak hours is shifted toward Off-peak hours without creating the peak on these hours. Above all this discussion if we look for scheduled load behavior for three price tariffs, load using RTP signals is much closer to objective load curve than other price tariffs.

Cuckoo search optimization technique for multi objective home energy management 30

400

40 Unscheduled Cuckoo Scheduled RTP

5

0

10

Electricity Cost (¢)

20

Load (kWh)

10

Electricity Price (¢)

Unscheduled Cuckoo Scheduled Objective Load RTP

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

200

0

Time (hours)

20

Electricity Price (¢)

15

7

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Time (hours)

(a) RTP

(a) RTP

15

20

250 Unscheduled Cuckoo Scheduled TOU

Unscheduled Cuckoo Scheduled Objective Load TOU

18

200

10

150 14

100 12

50

5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

0

Time (hours)

(b) TOU 140

Load (kWh)

100

8

80

6

60

4

40

2

20

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

(c) CPP Fig. 2 Load pattern per hour during a day

150 Unscheduled Cuckoo Scheduled CPP

120

10

Time (hours)

1500

Electricity Cost (¢)

Unscheduled Cuckoo Scheduled Objective Load CPP

Electricity Price (¢)

14

0

8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Time (hours)

(b) TOU

12

10

1000

100

500

50

0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Time (hours)

(c) CPP Fig. 3 Cost per hour during a day

Electricity Price (¢)

0

Electricity Price (¢)

5

Electricity Cost (¢)

Load (kWh)

15

Electricity Price (¢)

16 10

8

Adia Khalid et al. 2000

7

1800 6

5

1400 1200

4

PAR

Total Electricity Cost (¢)

1600

1000

3

800 600

2

400 1 200 0

0 Unscheduled

Cuckoo Scheduled

Unscheduled

(a) RTP

Cuckoo Scheduled

(a) RTP

1800

7

1600

6

5 1200 4

1000

PAR

Total Electricity Cost (¢)

1400

800

3

600 2 400 1

200 0

0 Unscheduled

Cuckoo Scheduled

Unscheduled

(b) TOU

7000

8

6000

7 6

5000

5 4000

PAR

Total Electricity Cost (¢)

(b) TOU

Cuckoo Scheduled

4

3000 3 2000

2

1000

1

0

0 Unscheduled

Cuckoo Scheduled

(c) CPP Fig. 4 Cost per hour during a day

Unscheduled

Cuckoo Scheduled

(c) CPP Fig. 5 Load pattern per hour during a day

Cuckoo search optimization technique for multi objective home energy management

9

Table 1 Appliances used in simulations Group

Appliances Vacuum cleaner Water heater Water pump Interruptible load Dish washer Iron Refrigerator AC Washing machine Non-interruptible load Cloth dryer Oven Blender Light1 Non-schedule able load Light2 Light3 Light4

Power Rate (kWh) 0.7 5 1 1.8 1 0.225 1.5 0.7 5 2.15 0.3 0.03 0.03 0.011 0.18

Daily Usage

Suggest Documents