Kỹ thuật và công nghệ phòng chống thiên tai, biến đổi khí hậu, lũ lụt

Genetic Algorithm Based Solar Tracking System D.F.Fam & S.P. Koh & S.K. Tiong & K.H. Chong Department of Electronic & Communication Engineering, Universiti Tenaga Nasional, Km 7, Jalan Kajang-Puchong, 43009 Kajang, Selangor. se20597@uniten.edu.my,johnnykoh@uniten.edu.my,siehkiong@uniten.edu.my,chongkh@uniten.edu.my Abstract: The current trend in solar concentrator tracking system is to use an open-loop local controller that computes the direction of the solar vector based on geographical location and time. But it is not accurate because it has error from computing the sun’s position, mechanical, controller systems and installation. Literature suggested that the photovoltaic panels could produce maximum power if the panels have angle of inclination zero degree to the sun position. In this research, genetic algorithm is one of the optimization techniques used to maximize the performance of solar tracking system. This work evaluates the best combination of GA parameters by always fine-tuning the position of solar tracking prototype to receive maximum solar radiation. Simulation results demonstrated the ability of GA solar to produce consistent result despite of different environmental conditions. Index Terms—genetic algorithm, solar tracking, photovoltaic panel 1. Introduction The solar tracker, a device that keeps PV or photo-thermal panels in an optimum position perpendicular to the solar radiation during daylight hours, increases the collected energy. The first tracker introduced by Finster in 1962, was completely mechanical. One year later, Saavedra presented a mechanism with an automatic electronic control, which was used to orient an Eppley pyrheliometer [1]Solar tracking can be implemented by using one- axis, and forhigher accuracy, two-axis sun- tracking systems. For a two-axis sun-tracking system, two types are known as: polar (equatorial) tracking and azimuth/elevation (altitude–azimuth) tracking.[2] High- concentration solar requires the sun to be tracked with great accuracy for maximum output voltage. The accuracy required depends on the specific characteristics of the concentrating system being analyzed. In general, the higher system concentration needs the higher accuracy tracking system. The current trend in solar concentrator tracking system is to use an open-loop local controller that computes the direction of the solar vector based on geographical location and time. But it is not enough accurate because it has error from computing the sun’s position, mechanical, controller systems and installation. [3]The abundance of solar energy throughout the whole year in Malaysia due to the geographic location near the Equator line provides strong reason for the implementation of an efficient PV energy system. Studies show that solar panels constitute a large portion (57%) of the total cost to install PV energy system [4] Since the purchase of solar panel is quite expensive, therefore research has been heavily invested by Ministry of Science, Technology and Innovation on few local universities to study on the implementation of PV technology as a renewable energy in Malaysia to replace coal and gas which form the primary resources to generate electricity by Tenaga Nasional Berhad.The conventional solar panel which is used to produce power is not maximized to its peak performance due to its static placement which limits the area of exposure to the sunlight[5]Abdallah et al. designed and constructed a two-axes, open loop,PLC controlled sun-tracking system. Their work principle is based on mathematical definition of surface position that is defined by two angles: the slope of the surface, and azimuth angle. The slope was considered to be equal as zenith angle of the sun. Two tracking motors, one for the joint rotating about the horizontal N–S axis and the other for the joint rotating about the vertical axis were used.The daylight divided into four intervals and during each of them the solar and motors speed were defined and programmed into PLC.They predicted that the power consumption to drive motors and control systems hardly exceeds 3% of power saved by the tracking system. Fig. 1 shows energy comparison between the tracker and the fixed surface inclined at 32 °. They concluded that the use of two-axes tracking surfaces results in an increase in total daily collection of about 41.34% as compared to that of a fixed one [6,7].Bingol et al. proposed, implemented and tested a microcontroller based two-axis solar tracking system. They used light dependent resistors (LDR) as sensors, stepper motors as actuators and a microcontroller. In addition, the system was connected to a PC via RS232 for sun position monitoring. A crystal with a frequency of 4 MHz was used as a clock signal generator for the microcontroller. The panel degree from vertical axis was fixed at 50 °. The experimental study for two solar collector panels, one stationary and the other rotary were employed in the test. Temperature of the panels versus time was measured with a minute interval and 50 data were captured. The angle of intervals was almost 5.2 °. A distinction of 9 ° between rotary and stationary panel was observed. This result verified that the rotary panel containing solar tracking system took more light density than the stationary panel[8] Fig 1 Energy comparison between tracking and fixed solar system [9] In this research, solar cell polycrystalline and GA has been used to fully maximize the performance of the solar tracking prototype. GA is used to overcome the current limitation of the method that had been used by other researchers where the best GA parameters will be chosen based on the intensity fitness function and both angular axis will be simulated to face the solar panel at the right angle for maximum power generation. In the following section, methodology will be discussed followed by results of the simulation and detailed discussion. Finally, conclusion is presented for further improvement in the future. 2. Methodology Methodology part is divided into few areas which include initial population, evaluation, selection, crossover and mutation. In this paper, the flowchart of the system development is highlighted as shown in Fig. 2 Fig. 2: Flowchart of the system development 2.1 Initial Population Initial population is randomly generated in 8 real values as follow: T1 = [ 0, 10, 11, 20, 1, 10,15,19] T2 = [ 1, 9, 15, 19, 1, 8,11,16] T3 = [ 0, 5, 14, 19, 1, 7,12,18] T4 = [ 1, 5, 13, 18, 1, 8,15,19] T5 = [ 0, 8, 13, 18, 1, 9,13,17] T6 = [ 1, 4, 11, 17, 1, 8,12,19] T7 = [ 1, 7, 15, 16, 0, 9,14,18] T8 = [ 1, 3, 14, 16, 1, 8,15,17] T9 = [ 1, 2, 12, 17, 1, 3,12,16] T10 = [ 1, 8, 12, 18, 1, 9,13,18] The first gene and fifth gene of each chromosome are used to indicate the direction of solar tracking axles, both horizontal and vertical axles to be left indicated by 0 or right indicated by 1. Second to fourth gene of each chromosome are used to indicate the different angles of horizontal axle to be positioned in order to get the maximum sun intensity. At the same time, sixth to eighth genes of each chromosome are used to indicate the different angles of vertical axle to be positioned in order to get the maximum sun intensity Both motors are used to control the horizontal and vertical axles which will be directed by PLC controller. 2.2 Evaluation In this wok, the objective function F(Tk) need to be evaluated by measuring the voltage from solar cell polycrystalline. Tk =(A^B) exp[-C X (dk )E] Where dk represents voltage from solar cell polycrystalline, A and B are fixed coefficient while C and E are tuning parameters. Each chromosome will be evaluated based on the objective function and fitness is simply equal to the value of objective function F (Tk) , k = 1,2…..k+1, where k = population size From the evaluation, strongest chromosome and weakest will be identified. Program Initialization Selection Start Evaluation Crossover Y Solution found End N Gen > Max Gen Y End N 2.3 Selection A roulette wheel selection approach is used in this research where it belongs to the fitness proportional selection and new population is selected with based on probability distribution according to fitness values. Fitness value, F (Tk ) will be calculated for each chromosome, (Tk ) Total fitness equation for the population will be as below: K+1 F = Σ F (Tk ) (1) K=1 For each chromosome, the probability equation, Pk Pk = (Tk ) , k = 1,2… K+1 where k= population size (2) F(Tk ) For each chromosome, cumulative probability equation, Ck K+1 Ck = Σ Pk k = 1,2… K+1 where k= population size (3) K=1 Roulette wheel is spanned for each selection process and a new single chromosome will be selected based on its fitness value. 2.4 Crossover Crossover used in this research is one cut point method which randomly select on cut point location and each side of chromosome will be exchanged between two parents to generate offspring. Consider two chromosomes as follow and cut point is selected after 3rd gene. T1 = [ 1, 10, 15, 20, 1, 9,11,19] T2 = [ 1, 9, 14, 18, 1, 8,13,17] The resulting offspring by exchanging the right part of their parents would be as below: T1 = [ 1, 10, 14, 20, 1, 9,11,19] T2 = [ 1, 9, 15, 18, 1, 8,13,17] The probability of crossover for each experiment is set as Pc= 0.8, therefore, 80% of chromosomes will undergo crossover. 2.5 Mutation Mutation changes one or more genes according to probability. Assume 3rd gene and 4th gene of the chromosome T1 is selected for a mutation. Since the gene is 15 and 20, therefore, both number will be interchanged and chromosome after mutation is as below : T1 = [ 1, 10, 15, 20, 1, 9,11,19] T1m = [ 1, 10,20,15, 1, 5,13,20] The probability of mutation for each experiment is set as Pc= 0.025, therefore, 2.5% of chromosomes will undergo mutation. After each chromosome have finished the cycle of crossover and mutation, the process will go back to evaluation procedure to continue the GA operations until the experiment ends. 3. Simulation A solar tracking has been developed to evaluate the application of genetic algorithm as depicted in Figure 3. It would explore the intensity of sunlight at different angles and locate the highest intensity with the GA simulation. The simulation has been carried out using the GA parameters as given in 3 tables below, Table 1, Table 2 and Table 3 with the objectives to study Table I: GA simulation parameter Simulation Parameter Value Maximum Generation Population, po Chromosome length Selection Method Crossover Rate, pc Mutation Rate, pm Mutation Point, mp No. of Best Chromosomes Kept, kb Crossover Type 50 10 8 Roulette Wheel 80% 0.025 2 1 Dynamic The solar tracking is placed at the origin point of (Xo=45 °, Yo=45 °). The default base point is at the centre of the workspace. In the simulation, the solar cell will keep on searching the highest intensity location with GA searching method. Both stepper motors controlling X and Y axis of solar tracking will receive the signals through motion controller to determine the angles of movement for both axis. Highest intensity that is absorbed by solar cell will convert the digital voltage to analogue signal to be transmitted to Visual basic program via Programmable logic controller Panasonic FPX-C14R. Figure 3 : A solar tracking has been developed to evaluate the application of genetic algorithm 4. Preliminary Results From the voltage measurement on the solar cell at different angles manually, the results which had taken on shiny day around 11am are shown in Table 2. Y AIXS (°) X AIXS (°) Result (V) 0 0 10.05 0 15 10.04 0 30 9.85 0 45 9.77 0 -15 9.78 0 -30 9.68 0 -45 9.43 15 0 9.64 15 15 9.84 15 30 9.81 15 45 9.74 15 -15 9.74 15 -30 9.44 15 -45 9.31 30 0 9.58 30 15 9.46 30 30 9.18 30 45 8.88 30 -15 9.43 30 -30 9.5 30 -45 9.53 45 0 9.82 45 15 9.77 45 30 9.71 45 45 9.59 45 -15 9.77 45 -30 9.66 45 -45 9.44 -15 0 9.86 -15 15 9.8 -15 30 9.78 -15 45 9.17 -15 -15 9.8 -15 -30 9.67 -15 -45 9.35 -30 0 9.77 -30 15 9.64 -30 30 9.42 -30 45 9.22 -30 -15 9.84 -30 -30 9.78 -30 -45 9.58 -45 0 8.93 -45 15 9.21 -45 30 9.32 -45 45 9.3 -45 -15 8.85 -45 -30 8.83 -45 -45 8.74 Table 2 : Voltage measurement corresponding to different angles X & Y 0 10 20 30 40 50 60 70 80 90 100 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 0.055 0.06 0.065 Generation Fi tn es s va lu e Best: 0.017065 Mean: 0.017066 Best fitness Mean fitness Graph 1 : Best Fitness Value- 0.017065 using GA 10 20 30 40 50 60 70 80 90 100 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 Generation Best, Worst, and Mean Scores Mean Score Best Score Worst Score Graph 3 : Best, worst and mean score for each generation using GA 10 20 30 40 50 60 70 80 90 100 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Generation A ve rg ae D is ta nc e Average Distance Between Individuals data1 Graph 2 : Average distance between individuals in each generation using GA 1 2 3 4 5 6 7 8 9 0 0.5 1 1.5 2 2.5 3 Selection Function Individual N um be r o f c hi ld re n state.Selection Graph 4 : Number of children that is produced by each individual using GA 0 10 20 30 40 50 60 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 Iteration Fu nc tio n va lu e Best Function Value: 0.017204 Graph 5 : Best Function Value- 0.01720 produced using Simulated Annealing 0 10 20 30 40 50 60 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 Iteration Fu nc tio n va lu e Best Function Value: 0.017072 Graph 7 : Best Function Value- 0.01707 produced using Threshold acceptance 0 10 20 30 40 50 60 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 Iteration Fu nc tio n va lu e Current Function Value: 0.01724 Graph 6 : Function Value for each iteration using simulated annealing 0 10 20 30 40 50 60 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 Iteration Fu nc tio n va lu e Current Function Value: 0.037029 Graph 8 : Function Value for each iteration using threshold acceptance 5. Discussion From the result, graph 1 shows the best fitness value 0.01706 that could be achieved using Genetic Algorithm through 50 generation. Graph 2 shows various average distances between each individual for 50 generation. Initial generation involves more activities in searching for optimum value which causes a larger distance split among each individual. Once the search space is narrowed down towards achieving global optimum value, distance between individuals is getting smaller and convergence moves to the vicinity of global minimum value, which is 22nd generation onwards till 50th generation indicated in graph 2. Graph 3 indicates best, worst and mean score for each generation where respective scores achieved by each fitness value during iteration is recorded. Graph 4 shows number of children that is produced by each parent along 50 generation. It is clear that fourth parent is having the highest tendency of producing highest off springs compared to other 9 parents. As compared to Graph 5, 6 and 7 and 8 where results are obtained using simulated annealing and threshold acceptance respectively, this is to prove that Genetic Algorithm is converging to the global minimum value by having lowest fitness value as shown below: Optimization Method Fitness Value Voltage Genetic Algorithm 0.01706 10.05 Simulated Annealing 0.01720 10.045 Threshold Acceptance 0.01707 10.050 6. Conclusion The proposed algorithm is used to control both X and Y motors so that solar tracking can be used to track the highest intensity. This experiment has been carried out in the outdoor working space to test the efficiency of this solar tracking. In this research, a simulator package has been developed and comparison between few other optimization methods has been done and best fitness value show that Genetic Algorithm performs better than Simulated Annealing and Threshold Acceptance. The proposed method improves search speed, good accuracy and approximate solution 7. Reference [1] Roth P, Georgiev A, Boudinov H. Cheap two-axis sun following device, Energy Conversion and Management 2005;46:1179–92 [2] Pitak Khlaichom Kawin Sonthipermpoon, Optimization of solar tracking system based on genetic algorithm, 3rd Conference of the Energy Network of Thailand,2007 [3] Hossein Mousazadeh,Alireza Keyhani, Arzhang Javadi, Hossein Mobli,Karen Abrinia,Ahmad Sharifi, 2009. A review of principle and sun-tracking methods for maximizing solar systems output, Renewable and Sustainable Energy Reviews 13 (2009) 1800-1818 [4] J. Enslin, “Renewable energy as an economic energy source for remote areas,” in Renewable Energy, vol. 1, pp. 243–248, 1991. [5] B. Koyuncu and K. Balasubramanian, “A microprocessor controlled automatic sun tracker,” IEEE Transactions on Consumer Electronics, vol. 37, no. 4, pp. 913–917, 1991 [6] Abdallah S, Nijmeh S. Two axes sun tracking system with PLC control. Energy Conversion and Management 2004;45:1931–9. [7] Mamlook R, Nijmeh S, Abdallah SM. A programmable logic controller to control two axis sun tracking system. InformationTechnology Journal