List of contents. List of Tables.III List of Figures IV List of Symbols and abbreviations... VI Abstract. VII CHAPTER INTRODUCTION...

Transcription

1 List of contents List of Tables.III List of Figures IV List of Symbols and abbreviations... VI Abstract. VII CHAPTER INTRODUCTION THESIS OBJECTIVES THESIS OUTLINE... 2 CHAPTER LITERATURE REVIEW INTRODUCTION RENEWABLE ENERGY FORMS SOLAR-PHOTOVOLTAIC ENERGY Growth of photovoltaic s Advantages of Photovoltaic systems Basic Types of PV cells Equivalent circuit and mathematical model Non Linear characteristics of PV s TYPES OF PHOTOVOLTAIC SYSTEMS Stand alone systems Grid-connected systems CONCLUSION CHAPTER MAXIMUM POWER POINT TRACKING INTRODUCTION DC-DC CONVERTERS Boost converters i

2 3.3 FAMOUS MPPT METHODS THE PERTURB AND OBSERVE METHOD MPPT USING ARTIFICIAL NEURAL NETWORKS Collecting Data Selecting Network Structure Training the network Testing the network HYSTERESIS CURRENT CONTROL OF INVERTER CONCLUSION CHAPTER CASE STUDIES AND SIMULATION RESULTS INTRODUCTION SYSTEM UNDER STUDY PV MODEL VERIFICATION BOOST CONVERTER MODEL NEURAL NETWORK CONTROLLER MODEL PERTURB AND OBSERVE CONTROLLER INVERTER CONTROL SIMULATION RESULTS CASE 1: System without MPPT CASE 2: System with MPPT using the proposed ANN method CASE 3: System with P&O MPPT algorithm ANN Vs P&O MPPT CONCLUSION CHAPTER CONCLUSION REFERENCES APPENDIX (A)- DATA SHEETS APPENDIX (B)- ANN TRAINING POINTS ii

3 APPENDIX (C)- MATLAB CODES APPENDIX (D)- BOOST CONVERTERS OPERATION List of Tables Table 4. 1 SIMULATION PARAMETERS Table 4. 2 KC200GT module parameters Table 4. 3 ANN test results iii

4 List of Figures Fig. 2. 1: Renewable energy share of global electricity production Fig. 2. 2: Parabolic trough systems using mirrors... 5 Fig. 2. 3: PV panels to generate electricity directly from the sunlight... 6 Fig. 2. 4: Process flow diagram of Bio-Refinery... 6 Fig. 2. 5: Dry steam power plant using geothermal energy... 7 Fig. 2. 6: Average annual growth rates of renewable energy capacity and bio-fuel s production... 8 Fig. 2. 7: Solar PV existing world capacity Fig. 2. 8: Mono-crystalline PV cell Fig. 2. 9: Poly-crystalline PV cell Fig : Amorphous PV cell Fig : PV module equivalent circuit Fig : Effect of temperature changes on I-V curves Fig : Effect of solar irradiance changes on I-V curves Fig : Effect of temperature changes on P-V curves Fig : Effect of solar irradiance changes on P-V curves Fig : Example of stand-alone PV system Fig : Example of Grid-connected PV system Fig. 3. 1: PV curve showing Maximum Power Point Fig. 3. 2: A complete Grid-Connected PV system Fig. 3. 3: Perturb and Observe algorithm flow chart Fig. 3. 4: Deviation of MPP with P&O method under rapid solar irradiance changes Fig. 3. 5: General Structure of a four inputs ANN Fig. 3. 6: Proposed ANN structure for MPPT Fig. 3. 7: Block diagram of the inverter control scheme Fig. 3. 8: Hysteresis current control block diagram Fig. 3. 9: Waveform of hysteresis switching for phase (a) Fig : A typical 3-phase inverter circuit diagram iv

5 Fig. 4. 1: The complete integrated system under study Fig. 4. 2: Simulation of module current in SIMULINK Fig. 4. 3: Simulation of reverse leakage and photovoltaic currents Fig. 4. 4: Simulation of the photovoltaic module Fig. 4. 5: I-V curves at different temperatures found in KC200GT data sheet Fig. 4. 6: I-V curves at different temperatures obtained by MATLAB Fig. 4. 7: I-V curves at different irradiances found in KC200GT data sheet Fig. 4. 8: I-V curves at different irradiances obtained by MATLAB Fig. 4. 9: Boost converter model Fig : ANN training with MATLAB NNET toolbox Fig : Generated SIMULINK two layers ANN blocks Fig : MPPT using P&O Fig : Control of Three phase inverter Fig : Detailed control circuit of inverter Fig : Hysteresis switching Fig : (a) Reference voltage (step) (b) DC-link voltage Fig : The complete integrated system on SIMULINK Fig : Two different operating conditions P-V curves Fig : System without MPPT (a) Temperature (b) Irradiance (c) PV output Power (d) PV terminal voltage Fig : System with ANN MPPT (a) Temperature (b) Irradiance (c) PV output Power (d) PV terminal voltage Fig : 1 and 2 (a) PV output Power (b) PV terminal voltage Fig : Output power of the array incase of P&O and without MPPT Fig : Oscillations around maximum power voltage incase of small step size Fig : Output power of the array incase of P&O and ANN MPPT v

6 List of Symbols and Abbreviations MPPT ANN HCC P&O a R s R p N s N p I m I pv V t N cs q k T I o I pvn K i K v G G n I scn V ocn D Cp Maximum power point tracking Artificial neural networks Hysteresis current control Perturb and observe Diode ideality constant Array series resistance Array parallel resistance Number of series modules Number of parallel modules Module current Photovoltaic current Thermal voltage of the array The number of cells connected in series The electron charge Boltzmann s constant Temperature of P-N junction the reverse leakage current Generated current at nominal operating conditions Current temperature coefficient Voltage temperature coefficient Solar irradiance Solar irradiance at nominal operating conditions Short circuit current of the module at nominal operating conditions Open circuit voltage of the module at nominal operating conditions DC-DC converter s duty cycle Step Value to be added or subtracted. ABSTRACT vi

7 Nowadays, renewable energy resources play an important role in replacing conventional fossil fuel energy resources. Photovoltaic energy is one of the very promising renewable energy resources which grew rapidly in the past few years. Photovoltaic s has one major problem and which is that with the variation of the operating conditions of the array, the voltage at which maximum power can be obtained from it also changes. In this thesis, a PV model is used to simulate actual PV arrays behavior, and then a Maximum Power Point tracking method using neural networks is proposed in order to control the DC-DC converter. An inverter Hysteresis Current Control scheme is also developed to maintain the DC-link voltage at a constant value which facilitates the maximum power point tracking process. Furthermore, the proposed artificial neural network technique is compared with the conventional perturb and observe maximum power point tracking method. A grid-connected complete photovoltaic model is generated to simulate the actual life case. Simulation results shows that the proposed artificial neural network maximum power point tracking method gives faster response than the conventional perturb and observe method under rapid variations of operating conditions. Also the proposed inverter control scheme shows fast dynamic response to step changes and that facilitates the maximum power point tracking process along with the grid connection. vii

8 viii

9 CHAPTER 1 INTRODUCTION 1.1 THESIS OBJECTIVES Renewable energy resources play an important role in electric power generation. There are various renewable resources which is used for electric power generation, such as solar energy, wind energy, geothermal etc. Solar Energy is a good choice for electric power generation, since the solar energy is directly converted into electrical energy by solar photovoltaic modules. These modules are made up of silicon cells. When many such cells are connected in series we get a solar PV module. The current rating of the modules increases when the area of the individual cells is increased, and vice versa. When many PV modules are connected in series and parallel combinations we get a solar PV array, which is suitable for obtaining higher power output. The applications of solar energy are increasing, and many researches are done to improve the materials and methods used to harness this power source. Main factors that affect the efficiency of the collection process are solar cell efficiency, intensity of source radiation and storage techniques. The efficiency of a solar cell is limited by materials used in solar cell manufacturing. It is particularly difficult to make considerable improvements in the performance of the cell, and hence restricts the efficiency of the overall collection process. Therefore, the increase of the intensity of radiation received from the sun is the most attainable method of improving the performance of solar power. There are two major approaches for maximizing power extraction in solar systems. They are sun tracking, maximum power point (MPP) tracking or both. Later on in this thesis, two MPP tracking techniques are studied and compared. The first technique is based on artificial neural networks and the second one is based on the P&O method. Also a complete grid connected scheme is proposed along with a DC-AC inverter control technique based on hysteresis current control. 1

10 1.2 THESIS OUTLINE This thesis proposes an ANN maximum power point tracking method along with a control scheme of an inverter to tie the PV arrays with the grid. The proposed ANN method is compared then with the conventional P&O method. The thesis consists of five chapters in which the MPPT problem is discussed in details and the proposed control schemes are fully explained. Chapter two makes a literature review on various types of renewable energy and especially on PV s and its types and equivalent circuits and characteristics. From which the MPPT problem originates. Chapter three discusses the MPPT problem in details and shows different MPPT methods. Two MPPT techniques are discussed in details in this chapter (P&O and the proposed ANN) and also the role of DC-DC converters and DC-AC inverters is explained. Chapter four shows the simulation results for the grid connected PV system using both P&O algorithm and the proposed ANN method, also the inverter control scheme is tested and the PV model is verified. All the simulations are made using MATLAB/SIMULINK computer software. Finally in chapter five, an overall conclusion is done and all the outcomes of the thesis are stated. 2

11 3

12 CHAPTER 2 LITERATURE REVIEW 2.1 INTRODUCTION As Traditional fossil fuels are expected to run out in the near future, the world started to rely on renewable energy as a clean, cheap and permanent substitute. Renewable energy is the energy which is available in natural resources such as sun, wind, tides and earth s crust heat. These resources are renewable and naturally replenish. Unlike other conventional energy resources, they have almost zero carbon emissions which decrease the global warming problem and green house effect phenomena. Apart from the harmful pollution that traditional energy resources do, fossil fuels reserves are decreasing rapidly, and that leads to the continuous increase in its price while renewable energy resources are permanent and free. In the past few years, renewable energy started to take serious steps on the way of replacing conventional fossil fuel energy production. Figure 2.1 shows the renewable energy share of global energy production in year 2010[1]. Fig. 2. 1: Renewable energy share of global electricity production

13 2.2 RENEWABLE ENERGY FORMS 1- Wind Power Wind turbines are used to harvest the energy available in airflows. Current day turbines rated power range from 600kW to 7.5MW (see appendix A.1). The output power of a wind turbine increases with the increase of the cube of wind speed, thus turbines are always installed in high altitudes and especially in places known for high wind speeds [2]. 2- Solar Power Harvesting the power of the sun can be done with two major ways, the first one is to collect the solar heat with mirrors and concentrate it on pipes to exchange the heat of the sun with a certain fluid and then it can be used to generate electricity [3], see Figure 2.2. The second way is to generate electricity directly from the falling sun rays using static photovoltaic cells [4] as shown in Figure 2.3. Fig. 2. 2: Parabolic trough systems using mirrors 5

14 Fig. 2. 3: PV panels to generate electricity directly from the sunlight 3- Hydropower Hydropower is achieved by converting the potential energy of water stored in dams and water in falling water falls into usable electrical energy through the use of water turbines. Another way is putting turbines in fast flowing rivers to produce electricity without building water reservoirs or dams [5]. 4-Biomass Plants capture the energy of the sun through the process of photosynthesis. Later on these plants under special circumstances can release that trapped energy. Many methods are applied to generate bio-products from the organic wastes, see Figure 2.4. This is not only solving the energy problem, but also keeping the environment cleaner by getting rid of these wastes [6]. Also from this bio refinery process we can get fertilizers for the soil. Anima Manure Anaerobic Bio-digester Ethanol Process Ethanol Fuels Green Waste Gassifier Process Gassifier Process Hydrogen Gas Plastic Waste POET Process Gas/Oil Distillation Gasoline/Diesel Fig. 2. 4: Process flow diagram of Bio-Refinery For Electric Power 6

15 5-Geothermal Geothermal electricity is the electricity generated from geothermal energy. Technologies in use include dry steam power plants, flash steam power plants and binary cycle power plants, see Figure 2.5. Geothermal electricity generation is currently used only in 24 countries while geothermal heating is in use in 70 countries. Current world installed capacity is 11 GW with the largest capacity in the United States, Philippines and Indonesia [1]. Geothermal power is considered to be sustainable because the heat extraction is small compared to the Earth's heat content. The emission intensity of existing geothermal electric plants is on average 122 kg of CO2 per megawatt-hour (MW h) of electricity, which is a small fraction of that of conventional fossil fuel plants [7]. Fig. 2. 5: Dry steam power plant using geothermal energy 5-Wave Power Wave power is the transport of energy by ocean surface waves, and the capture of that energy using special machines, do useful work (electricity generation, water desalination, or the pumping of water into reservoirs). Machinery able to exploit wave power is generally known as a wave energy converter (WEC). Wave power is different from the tidal power. Wave power is currently used in Portugal and the United Kingdom. 7

16 2.3 SOLAR-PHOTOVOLTAIC ENERGY Growth of photovoltaic s As Hydro and wind powers are limited to certain geographic conditions, solar power started to take over in the past few years as the solar energy is generously spread across our planet. Photovoltaics are spreading in a very fast way around the globe as it is an easy and fast method of producing electricity through solar energy. Figure 2.6 shows the average annual growth rates of renewable energy, its obvious that photovoltaic has the highest growth rate among all other renewable energy resources in While one can notice that the world capacity of photovoltaic s almost doubled in 2010 as shown in Figure 2.7. Fig. 2. 6: Average annual growth rates of renewable energy capacity and bio-fuel s production [1]. Fig. 2. 7: Solar PV existing world capacity [1]. 8

17 2.3.2 Advantages of Photovoltaic systems The advantages of photovoltaic [4] systems are: 1- PV systems are considered static electricity generators as they create electricity directly from sunlight. They come prepackaged, ready to be mounted and wired. Modules contain no moving parts, eliminating service and maintenance needs. 2- PV systems come in a range of sizes and output suitable for different applications. They are lightweight, allowing for easy and safe transportation. 3- PV systems can be easily expanded by adding more modules either in series to expand the system s voltage or in parallel to enlarge the current. 4- PV systems are manufactured to withstand the most rugged conditions. Modules are designed to endure extreme temperatures, at any elevation, in high winds, and with any degree of moisture or salt in the atmosphere. Systems can be designed with storage capabilities to provide consistent, high-quality power even when the sun isn t shining. 5- PV systems cause no noise or carbon emissions i.e. no pollution. The disadvantages of photovoltaic systems are: 1- Very high manufacturing cost compared to other renewable resources. 2- Maximum power point problem. 3- Requires regular cleaning of its outer surfaces from dust. 4- Significantly low in efficiency Basic Types of PV cells Photovoltaic cells are manufactured in different forms; each has its uses and benefits compared to others. The most famous PV types are [4]: 1-Mono-crystalline (single crystalline) cells: Mono-crystalline cells are cut from a single crystal of silicon, see Figure 2.8. They are basically a slice of crystal which makes them very smooth in texture. Mono-crystalline cells are the most efficient, but also the most expensive to produce. 2-Poly-crystalline (multi-crystalline) cells: 9

18 Polycrystalline cells are made from a slice cut from a block of silicon, unlike Monothese cells consist of a large number of crystals. (See Figure 2.9). Photovoltaic solar panels made from these types of cell are slightly less efficient but also slightly cheaper than mono-crystalline cells. 3-Amorphous cells: Amorphous cells are manufactured by placing a thin film of amorphous (non crystalline) silicon onto a wide range of surfaces.(see Figure 2.10). Amorphous cells are the least efficient type of Photovoltaic solar panels but also the cheapest. They form flexible PV panels. Fig. 2. 8: Mono-crystalline PV cell. Fig. 2. 9: Poly-crystalline PV cell. 10

19 Fig : Amorphous PV cell Equivalent circuit and mathematical model A current source type PV model is discussed in this section [8]. The equivalent circuit is shown in Figure R N s s N p I I m R N s p N p V Fig : PV module equivalent circuit. Where R s is the array series resistance, R p is the array parallel resistance, N s and N p are the number of series and parallel modules respectively, I and V are the output current and voltage of the array and I m is the module current and can be obtained from the following equation: 11

20 N s V R s I N p I m I pvn p I 0 N p exp 1 (2.1) Vt an s Where, a is the diode ideality constant, V t is the thermal voltage of the array and can be obtained from the equation: V t N cs q kt (2.2) N cs is the number of cells connected in series, q is the electron charge, k is Boltzmann s constant and T is the temperature of the P-N junction in Kelvin s. Ipv is the photovoltaic current and can be expressed by: I pv I pvn K T i G G n (2.3) And Io is the reverse leakage current of the diode and can be calculated from: I 0 I V exp scn ocn K K av t i T v T 1 (2.4) Where: I pvn is the generated current at 25 o C and 1000W/m 2 (nominal conditions), K i, K v the current and voltage temperature confidents respectively, G is the irradiance and G n is the irradiance at nominal conditions, I scn,v ocn are the short circuit current and open circuit voltage respectively at nominal conditions and T is the difference between the actual and the nominal temperatures in Kelvin s. 12

21 2.3.5 Non Linear characteristics of PV s Photovoltaics have non linear characteristics, where the performance and output power are directly affected with the change of the operating conditions (temperature and solar irradiance. Figures 2.12, 2.13, 2.14 and 2.15 show the effect of changing the temperature and solar irradiance on PV s output current, voltage and power. Fig : Effect of temperature changes on I-V curves. Fig : Effect of solar irradiance changes on I-V curves. 13

22 Fig : Effect of temperature changes on P-V curves. Fig : Effect of solar irradiance changes on P-V curves. It is clear from the previous figures that the output power of PV s is directly proportional with the amount of solar irradiance falling on it, and inversely proportional with its temperature. Figures 2.14 and 2.15 show that with the change of the temperature and the solar irradiance the point at which maximum power can be obtained also changes, this means that the array terminal voltage must be varied using DC-DC converters in 14

23 order to track the maximum power point. Maximum power point tracking methods will be discussed in details in chapter TYPES OF PHOTOVOLTAIC SYSTEMS Stand alone systems This type is particularly appropriate for remote areas where basic energy requirements are limited, provided that a reasonable amount of solar insulation is available. Stand alone systems mainly consist of a PV panel, a battery bank for storage and an inverter for DC to AC conversion [9][10]. (See Figure 2.16). It is also possible to couple a PV system with a fossil-fuel general. In this type of system, the generator is used to recharge the PV battery during long periods of cloudy weather. This Hybrid system requires much less fuel and maintenance for the generator, while extending battery life. Fig : Example of stand-alone PV system. 15

24 2.4.2 Grid-connected systems Unlike stand-alone systems, utility-interactive systems are connected to the power line, as shown in Figure This system has PV solar modules which supply electrical power to the equipment though a high quality inverter. This inverter converts PV-generated DC to high quality AC normally available from the power company [11]. Also Electric Power stations based on PV modules are considered grid-connected systems as they generate huge amount of power and adds it to the grid power. Fig : Example of Grid-connected PV system. Grid-connected systems show recently very good potential compared to stand-alone ones, as grid-connected users can sell their unused extra electricity to the utility with high prices and still supply their needs at nights from the utility if they have shortage. 2.5 CONCLUSION In this chapter, an overview of the importance of renewable energy and its various types and resources is made. The Photovoltaic energy in particular is reviewed with its global growth, advantages, cell types and mathematical model along with the equivalent circuit. The two main PV system types are also discussed and their components are mentioned. 16

25 17

26 CHAPTER 3 MAXIMUM POWER POINT TRACKING 3.1 INTRODUCTION As stated before in Chapter 2, the maximum power point of any PV varies with the variation of the atmospheric conditions (solar irradiance and temperature). This means that there is always one optimum terminal voltage for the PV array to operate at with each condition as shown in Figure 3.1, to obtain the maximum power out of it i.e. increase the array s efficiency. Fig. 3. 1: PV curve showing Maximum Power Point. DC-DC converters play an important role in the maximum power point tracking process. As by connecting the array s output terminals with the DC-DC converter s input 18

27 terminals, the array voltage can be controlled by varying the duty cycle of the converter and the voltage at which maximum power is obtained can be maintained. DC-AC inverter s main task is to convert the DC electricity to AC and hence it could be tied to the grid. The inverter has also a very important role in the MPPT process which is fixing the DC-link voltage at certain value. As varying the duty cycle of the DC-DC converter will change the array terminal voltage (the DC-DC converter s input voltage) only in case of fixing the output voltage of the DC-DC converter at a certain value, a control scheme for the DC-AC inverter is proposed to keep the DC link voltage constant (which is the DC-DC output voltage also) and thus, varying the duty cycle of the DC-DC converter varies the array terminal voltage. The complete system discussed in this thesis is shown in Figure 3.2. PV Array DC/DC Converter DC/AC Inverter Grid Fig. 3. 2: A complete Grid-Connected PV system. 3.2 DC-DC CONVERTERS DC-DC converters have wide applications in PV systems. Whether it is boost converter [12][13], buck-boost converters [14][15][16] or Buck converters [17]. DC-DC converters are considered the main element in the maximum power point tracking process and without it the maximum power could not be achieved. In this thesis boost converter is used change the terminal voltage of the PV array and from which maximum power point tracking can be obtained. 19

28 3.2.1 Boost converters The maximum power point tracking is basically a load matching problem. In order to change the input resistance of the panel to match the load resistance, a DC to DC converter is required. (see appendix D for details on boost converter s theory of operation) The main equation of the boost converter is: V V i 1 1 D 0 (3.1) Where, V i is the input voltage to the boost converter V 0 is its output voltage, and D is the duty cycle. In the case of this thesis, V 0 is fixed using the inverter control scheme. And V i is at the same time the array terminal voltage which is controlled by varying the duty cycle D. 3.3 FAMOUS MPPT METHODS Conventional methods: The most famous conventional MPPT methods are the Perturb and Observe (Hill climbing) method [18][19], the Incremental Conductance method[20]-[23], the Fractional open circuit voltage method[24][25] and the Fractional short circuit Current method[26][27]. Artificial intelligence methods: There are two main AI methods, Fuzzy logic based MPPT [28] and Neural Networks based ones [29]. 20

29 Both Conventional and artificial intelligence methods has their advantages and drawbacks. Conventional methods are famous for their easy implementation and compatibility to operate with any Photovoltaic array, While they re disadvantages is that they are considered relatively slower than the artificial intelligence methods and not only that they show slow response in sudden temperature and solar irradiance changes, but also they may fail in tracking the maximum power point [30]. On the other hand, artificial intelligence methods show very fast response under any operating condition changes, give very accurate results and they are able to work under instant temperature or solar irradiance changes efficiently. The drawbacks of the AI methods that they are complicated in design, they need very fast processors to be implemented physically or otherwise they will run very slowly. For each PV array type, a separate model should be designed to guarantee that it will perform well which is considered also a disadvantage [30]. 3.4 THE PERTURB AND OBSERVE METHOD The Perturb and Observe method (P&O), sometimes called Hill climbing method, is the most famous MPPT technique. P&O is widely used as it is the simplest method among all MPPT ones. P&O is simply measuring the PV s terminal voltage and output current, from which the actual Power can be calculated and varying the duty cycle of the DC-DC converter is done until the MPP is achieved. As the name of the P&O method states, the process starts by operating the DC-DC converter with the initial set duty cycle, and then starts increasing the duty cycle with a certain step width (user defined), and the Power is observed with the addition of each step. If at a certain point the Power gets less than its previous value that means that the duty cycle should get one step in the opposite direction i.e. getting to the MPP again and etc (see Figure 3.3) 21

30 Begin P & O Measure: V(n), I(n) ΔVref (n) = Vref (n) Vref (n-1) P(n) = V(n) x I(n) ΔP(n) = P(n) - P(n-1) No ΔP(n) > 0? Yes Yes ΔV ref (n) > 0 No No ΔVref (n)>0 Yes?? Vref (n+1) = Vref (n+1) = Vref (n+1) = Vref (n+1) = Vref (n) - Cp Vref (n) + Cp Vref (n) - Cp Vref (n) + Cp Fig. 3. 3: Perturb and Observe algorithm flow chart. 22

31 With this algorithm the operating voltage V is perturbed with every MPPT cycle. As soon as the MPP is reached, V will oscillate around the ideal operating voltage. This causes a power loss which depends on the step width of a single perturbation i.e. the larger the step is, the larger the oscillations around voltage of maximum power is and vice versa [31]. If the step width is large, the MPPT algorithm will be responding quickly to sudden changes in operating conditions with the tradeoff of increased losses under stable or slowly changing conditions. If the step width is very small the losses under stable or slowly changing conditions will be reduced, but the system will be only able to respond very slowly to rapid changes in temperature and solar irradiance. The value of ideal step width is system dependant and needs to be experimentally determined. One drawback of the P&O algorithm is in case of sudden increase in the solar irradiance, the P&O reacts as if the increase occurred as a result of the previous perturbation of the array operating voltage [30][31]. The next operation, therefore, will be in the same direction as the previous one which may be the opposite direction of maximum power. Figure shows that the continuous perturbation in one direction will lead to an operating voltage away from the MPP voltage. When the irradiance change decreases or stops the MPPT will get back to its normal behavior. (see Figure 3.4) 23

32 Fig. 3. 4: Deviation of MPP with P&O method under rapid solar irradiance changes. 3.5 MPPT USING ARTIFICIAL NEURAL NETWORKS Artificial Neural Network (ANN) is an artificial network that mimics the human biological neural networks behavior, widely used in modeling complex relationships between inputs and outputs in nonlinear systems. ANN can be defined as parallel distributed information processing structure consisting of inputs, and at least one hidden layer and one output layer. These layers have processing elements called neurons interconnected together. (see Figure 3.5). 24

33 Fig. 3. 5: General Structure of a four inputs ANN. The main advantages of using the artificial neural network ANN controller are: 1. A neural network can perform tasks that a linear program cannot i.e. very suitable for non-linear systems. 2. When an element of the neural network fails, it can continue without any problem by their parallel nature. 3. A neural network learns and does not need to be reprogrammed. 4. It can be implemented in various applications. The main Disadvantages of using the artificial neural network ANN controller are: 1. The neural network needs training to operate. 2. The architecture of a neural network is sometimes complicated in design. 3. Requires high processing time for large neural networks. 25

34 An ANN is developed, such that the current solar irradiance and temperature are its inputs and the voltage, which corresponds to maximum power output of the array. The detailed ANN structure and data will be discussed in details in the following sections Collecting Data The first step in designing an ANN is to collect historical data on the problem that is being solved using the network. In case of MPPT lots of array solar irradiances and temperatures and their corresponding maximum power point voltages are required to in order to train the network. This obtained data is called (training points). The obtained training points for this thesis model is shown in Appendix B Selecting Network Structure As mentioned before, neural networks consist of a minimum of two layers (one hidden layer and another output layer). The input information is connected to the hidden layers through weighted connections where the output data is calculated. The number of hidden layers and the number of neurons in each layer controls the performance of the network. Until now, there are no guidelines for deciding a way to choose the number of neurons along with number of hidden layers for a given problem to give the best performance. And it is still a trial and error design method. The developed ANN in this thesis is a two inputs ( solar irradiance and Temperature ) with two layers ( one hidden layer and one output layer ) the hidden layer has ten neurons with tan-sigmoid activation function and the output layer has only one neuron with purelinear activation function which is the voltage at maximum power point. Figure 3.6 shows the discussed structure. 26

35 Inputs Hidden layer Output layer N(1) Irradiance N(2) Vmpp Temperature N(10) Fig. 3. 6: Proposed ANN structure for MPPT Training the network The collected training points are passed into the designed network in order to teach it how to perform when different points than the training points are inserted to it. A computer software are usually used to do this process. In this Thesis MATLAB/M-File is used to train the network (see appendix C.1) Testing the network Some of the collected Test points are kept as test points. The function of test points is to test the performance of the designed ANN after its training is finished as these test points will be new to it and thus we can judge whether it gives accurate results or not. In Chapter 4 some of the test points will be applied in order to find out how accurate the developed network is. (see appendix C.3). 27

36 3.6 HYSTERESIS CURRENT CONTROL OF INVERTER As previously in this chapter, the main function of the inverter is to interface the PV array with the grid. In the same time the inverter is used to maintain the voltage at the output side of the boost converter (the inverter s DC link) constant. In order to obtain this Hysteresis current control scheme (HCC) is used [32]. The control scheme is applied by taking a reference DC voltage signal and subtracting from it the actual DC voltage signal. The generated error signal is passed through a Proportional Integral controller (PI) which produces the direct current component i d. The quadrature current component i q is set to zero to prevent flow of reactive power from the grid Figure 3.7. Fig. 3. 7: Block diagram of the inverter control scheme. The park transformation is the used to convert the currents id and iq from synchronously rotating frame into the three phase currents through the following equation: i i i a( ref ) b( ref ) c( ref ) cos 2 sin 3 2 sin i cos i d q (3.2) 28

37 Where is the instantaneous angle calculated using the phased locked loop (PLL) circuit. The main function of the phased locked loop circuit is to make the parks transformation synchronized with the grid voltages [32]. The calculated reference currents are to be compared to the actual three phase grid currents using hysteresis comparators and the error signal which controls the switches of the inverter as shown in Figure 3.8. Hysteresis Ia (ref) + Ia (actual) - Ib (ref) + Hysteresis To inverter switches Ib (actual) - Hysteresis Ic (ref) + Ic (actual) - Fig. 3. 8: Hysteresis current control block diagram. Hysteresis comparators work on forcing the actual current in a certain band. For example in phase (a), when the actual current reaches the upper band, switch S4 will be on to force the current down, and when the current reaches the lower band, switch S1 will be on to force the current up and etc... The actual current is forced to stay within the band even if the reference current changes. The same happens with the other two phases. (see Figure 3.9 and 3.10). 29

38 Fig. 3. 9: Waveform of hysteresis switching for phase (a). Fig : A typical 3-phase inverter circuit diagram. Advantages of Hysteresis Current Control technique[32][33]: 1- Fast dynamic response. 2- Simple in implementation. 30

39 3- Hysteresis current control has rapid convergence for large scale current errors. Disadvantages of Hysteresis Current Control technique: 1- Variation of switching frequency. 2- Current Ripples. 3.7 CONCLUSION In this chapter, the maximum power point tracking problem is discussed, and the various types of DC-DC converters, which is the main tool used for obtaining the maximum power are mentioned. The boost converter theory of operation is also discussed in details. Further on, different famous MPPT methods are mentioned and their advantages and disadvantages are also highlighted, and then the two main methods used in this thesis the P&O and ANN are discussed in more details along with the control scheme of the DC- AC inverter. 31

40 32

41 CHAPTER 4 CASE STUDIES AND SIMULATION RESULTS 4.1 INTRODUCTION This chapter shows all the simulations for the Photovoltaic array, the boost converter, the proposed artificial neural network, Perturb and ovbserve algorithm and the proposed inverter control scheme. All the simulations are done using MATLAB/SIMULINK software. The simulation results and some case studies for the whole proposed system are discussed in this chapter. 4.2 SYSTEM UNDER STUDY The complete system is to be simulated using the MATLAB/SIMULINK (as shown in Figure 4.1), and by varying the operating conditions (solar irradiance and temperature), both conventional P&O algorithm and artificial neural networks MPPT methods will be compared. In addition to the MPPT techniques, the proposed inverter hysteresis current control scheme is tested under step changes of the DC reference voltage. The PV array is composed of (15x2) series and parallel modules respectively with a total output power of 6kW. PV Array DC/DC Converter DC/AC Inverter Grid Irradiance Temp PWM Vdc Hysteresis Control I grid ANN MPP Calculation Vref Parks Transformation V grid Fig. 4. 1: The complete integrated system under study using ANN. 33

42 Table 4. 1 SIMULATION PARAMETERS Shows the simulation parameters for the proposed system Quantity Grid voltage Frequency Switching frequency DC link capacitor C Value 440V 60 Hz 1kHz 1mF DC link Voltage 400V Converter inductance H Converter Capacitor 0.04F Sampling period 1µS 4.3 PV MODEL VERIFICATION The PV mathematical model discussed in chapter 2 section is modeled and simulated using the MATLAB/ SIMULINK. The previously discussed module current (chapter 2, equation 2.1) SIMULINK model can be seen in Figure 4.2, while the reverse leakage and the photovoltaic currents (discussed in chapter 2, equations 2.4 and 2.3 respectively) simulation are in Figure 4.3. Fig. 4. 2: Simulation of module current in SIMULINK. 34

43 Fig. 4. 3: Simulation of reverse leakage and photovoltaic currents. The MATLAB/SIMULINK model of the simulated PV is shown in Figure 4.4 Fig. 4. 4: Simulation of the photovoltaic module. The MATLAB/SIMULINK model is tested by inserting all the required data shown in Table 4-2 to simulate the KYOCERA KC200GT module. The data sheet [34] of the simulated module is shown in appendix A.2. By comparing the KC200GT datasheet I-V curves at different values of solar irradiance and temperature with the one s obtained by the MATLAB model, shown in Figures 4.5, 4.6, 4.7 and 4.8, It is very clear that the curves obtained by the model are almost identical to the one s found in the module data sheet which proves that the model is reliable. 35

44 Table 4. 2 KC200GT module parameters Quantity Value I max power 7.61A V max power 26.3V P max W I short circuit 8.21A V open circuit 32.9V I leakage 9.825x10-8 A I photovoltaic 8.211A Diode ideality constant 1.3 (a) Parallel resistance Ω Series resistance 0.221Ω Fig. 4. 5: I-V curves at different temperatures found in KC200GT data sheet. 36

45 Current w/m C 50C C Voltage Fig. 4. 6: I-V curves at different temperatures obtained by MATLAB. Fig. 4. 7: I-V curves at different irradiances found in KC200GT data sheet. 37

46 Current W/m2 800 W/m2 600 W/m2 400 W/m2 200 W/m2 Temperature=25C Voltage Fig. 4. 8: I-V curves at different irradiances obtained by MATLAB. 4.4 BOOST CONVERTER MODEL The MATLAB/SIMULINK model of the boost converter is shown in Figure 4.9. The boost converter plays very important role as it varies the PV array terminal voltage with the change of the duty cycle. The duty cycle will be determined depending on the signal of the maximum power point tracker whether it is P&O or ANN as it is discussed in the following sections. Fig. 4. 9: Boost converter model. 38

47 4.5 NEURAL NETWORK CONTROLLER MODEL As stated before in chapter 3, training points should be obtained in order to start our work with any ANN. The training points are obtained through varying the irradiance and temperature of the array and taking values of voltage and currents and maximum power. The training points obtained are shown in appendix B.1. ANN is formed using MATLAB M-FILE and the code is shown in appendix C.2. The proposed ANN has two inputs (solar irradiance and temperature) and one output which is the voltage at maximum power point. The network is trained using the MATLAB NNET tool box [35], see Figure 4.10 Fig : ANN training with MATLAB NNET toolbox. This M-FILE is then transformed to SIMULINK blocks using the command (gensim) which facilitates the usage of the designed ANN with the whole SIMULINK system and the output of the ANN is used to calculate the duty cycle for MPPT, see Figure

48 Fig : Generated SIMULINK two layers ANN blocks. The generated ANN is then tested using some test points and the results can be shown in Table 4-3. Table 4. 3 ANN test results Test Point Temperature Irradiance Vmpp Vmpp Pmax Pmax o C W/m 2 actual ANN actual ANN 1 18 o C V V 6208W 6206W 2 27 o C V V W W 3 38 o C V V 5322W 5331W It is obvious that the ANN results are very close to the actual one s which implies the high accuracy of the designed network. 4.6 PERTURB AND OBSERVE CONTROLLER The Perturb and Observe algorithm discussed before in chapter 3 section 3.4 is constructed using MATLAB M-FILE code within SIMULINK, the code is shown in appendix C.4 and its output is connected to the Boost converter to achieve MPPT. Figure 4.12 show the SIMULINK model of the MPPT using P&O method, 40

49 Fig : MPPT using P&O. 4.7 INVERTER CONTROL Hysteresis current control is performed on the three phase inverter to keep the DClink voltage constant at any required value. The whole process discussed before in chapter 3 section 3.6 is simulated on SIMULINK, see Figures 4.13 and The hysteresis comparison process is shown in Figure Fig : Control of Three phase inverter. 41

50 Fig : Detailed control circuit of inverter. Fig : Hysteresis switching. The controller is tested under step reference voltage (starts with 400V and then drops to 300V after 0.1 seconds). Figure 4.16 shows the dynamic results of the inverter control which shows fast dynamic response and very accurate results and those as discussed before in chapter 3 are what characterize the hysteresis current control among different control topologies of the inverter. 42

51 DC voltage Voltage (reference) The inverter succeeded to keep the DC link voltage at a constant value which makes the DC-DC converter s task of MPPT much easier. At the same time the inverter converts the DC power into AC and that is its main task so that the PV system could be easily tied with the grid (a) (b) Time(sec.) Fig : (a) Reference voltage (step) (b) DC-link voltage. 4.8 SIMULATION RESULTS All the discussed elements in this chapter are to be bound together to form a complete integrated grid-connected PV system as seen in Figure This system is to be studied in different cases; without MPPT, with the artificial neural network controller, and with perturb and observe controller and then comparisons are made. The inverter control is set to 400V for the DC-link. A PV module KC200GT parameters are inserted into the PV model [34] and two P-V curves are obtained at two different operating conditions (temperature 43

52 Power(watt) and irradiance) as shown in Figure An array of 15 by 2 series and parallel modules respectively is formed with a total output power 6004W at nominal operating conditions. Figure 4.18 shows that for each operating conditions there is a different voltage to operate at in order to obtain the maximum power out of the array. Fig : The complete integrated system on SIMULINK C & 1175 w/m2 25C & 1000 w/m v 395v Voltage (volts) Fig : Two different operating conditions P-V curves. 44

53 Voltage(volt) Power(watt) Irradiance Temperature CASE 1: System without MPPT First the system is operated at certain duty cycle, which achieves the maximum output power at nominal operating conditions (25 o C and 1000W/m 2 ) and then the temperature and irradiance are to be changed from 25 and 1000 respectively to 33 and Figure 4.19 shows that power output after the conditions have changed is not the maximum power as the PV array voltage remains 395V which is not the voltage required to achieve the maximum power in that case ( it should be 375V) (a) (b) (c) (d) Time(sec.) Fig : System without MPPT (a) Temperature (b) Irradiance (c) PV output Power (d) PV terminal voltage. It is clear that when operating conditions changes, the duty cycle of the DC-DC converter should also change to compensate the difference in the voltage i.e. track the maximum power. The power changes with the conditions from 6004W to 6670W while the curves in Figure 4.19 shows that the maximum power in the new operating conditions is 6775W i.e. there is a loss of 105W. 45

54 Voltage(volt) Power(watt) Irradiance Temperature CASE 2: System with MPPT using the proposed ANN method For the same conditions, the system is operated again but with the proposed artificial neural network controller to control the duty cycle of the Boost converter. Figure 4.20 shows the results in this case (a) (b) (c) (d) Time(sec.) Fig : System with ANN MPPT (a) Temperature (b) Irradiance (c) PV output Power (d) PV terminal voltage. The artificial neural network controller succeeds in changing the duty cycle of the boost converter to make the PV array s terminal voltage 375V which makes the output power of the array at its maximum value 6675W. Figure 4.21 shows the cases 1 and 2 on the same curve. 46

55 Voltage(volt) Power(watt) with MPPT with out MPPT (a) with MPPT with out MPPT (b) Time(sec.) Fig : 1 and 2 (a) PV output Power (b) PV terminal voltage CASE 3: System with P&O MPPT algorithm Using the previously discussed P&O in chapter 3 method, the P&O algorithm is done using MATLAB code as shown in appendix(c). The step size is set to 1 volt (which is equivalent to (2.5x10-3 ) change in the duty cycle. The increase of step size would lead to fast response but less accuracy, and the decrease of step size would make the system very slow thus the step size must be set to a suitable value. (The same operating condition changes applied in case 1 and 2 are applied in this case). Figure 4.22 shows system with P&O method versus system without any MPPT while in Figure 4.23 it is obvious that the P&O algorithm keeps perturbing around the voltage at maximum power by adding and subtracting the selected step size. 47

56 Power(watt) with P&O MPPT with out MPPT Fig : Output power of the array incase of P&O and without MPPT Fig : Oscillations around maximum power voltage incase of small step size ANN Vs P&O MPPT Figure 4.24 shows the proposed ANN method versus the P&O method, it s clear that if the step size of P&O is well adjusted MPP will be achieved, but still the ANN is very fast compared to P&O as there is no perturbation and the voltage at maximum power is estimated instantaneously i.e. the DC-DC converter s duty cycle changes at no time and keeps its new value until the next operating condition change. 48

57 Power(watt) with ANN MPPT with P&O MPPT Fig : Output power of the array incase of P&O and ANN MPPT. 4.9 CONCLUSION In this chapter, simulation is made using MATLAB/SIMULINK software for a complete grid-connected PV system. Each component is simulated and discussed in details. The PV model was verified and it gave almost typical results like the ones supplied by the manufacturer data sheet. The two MPPT methods are developed (P&O and artificial neural networks) and are compared together. The proposed neural network method showed faster response than the P&O and gave accurate results. Also the inverter control scheme was tested under step changes and showed fast dynamic response and it could keep the DC-link s voltage constant at any desired value accurately. 49

58 50

59 CHAPTER 5 CONCLUSION 5.1 CONCLUSION This chapter summarizes the main points presented in this work with suggested future research on the proposed method of maximum power point tracking. This thesis has presented a literature review on the famous renewable energy resources and then highlighted the photovoltaic energy and its types and characteristics which is the main topic of this thesis. An artificial intelligent maximum power point tracking technique using neural networks is proposed, which predicts the appropriate duty cycle for which the DC-DC converter can operate with and thus maximum power can be obtained from the PV system. The system is tied to the grid with a DC-AC inverter. A PV simulator is also developed using MATLAB/SIMULINK software which if supplied by the required data can simulate the behavior of any PV array. Also DC-DC boost converter model is simulated which is the key for changing the PV s terminal voltage to track the maximum power. DC-AC inverter control scheme is developed using hysteresis switching technique which gives fast dynamic response and accurate results. This scheme keeps the DC link s voltage constant at any desired value and that facilitates the role of the DC-DC converter in maximum power tracking. Also a complete integrated grid-connected scheme was proposed to simulate the real time cases. The conventional Perturb and Observe MPPT algorithm is developed through MATLAB code and tested and compared with the artificial neural network MPPT method under sudden temperature and irradiance changes and the ANN method gave very fast and accurate response. 51

60 5.2 FUTURE WORK 1- Implementation of a physical model for the artificial neural network MPPT technique using microcontrollers and testing it on a real PV panel. 2- Comparing between different MPPT techniques and the proposed neural network one. 3- Studying the effect of the hysteresis current control method and its effect on power quality from the utility grid point of view. 52

61 53

62 REFERENCES [1] REN21 Renewables 2011 Global Statues Report. [2] T.Burton, D.Sharpe, N.Jenkins and E.Bossanyi (2001) wind energy handbook. England: Wiley. [3] Soteris Kalogirou.(2009), Solar Energy Engineering: Processes and Systems. Academic press. [4] Adolf Goetzberger.(2005) Photovoltaic Solar Energy Generation. Springer. [5] Francesco Carrasco. (2011) Introduction to Hydropower. The English press. [6] Elizabeth Hood. (2011) Plant Biomass Conversion. England : Wiley. [7] Val Pierce.(2011) Introduction to geothermal power. England: the English press. [8] M.G.Villalva, J.R.Gazoli and E. Ruppert F. Comprehensive Approach to Modeling and Simulation of Photovoltaic Arrays. IEEE Transactions on power electronics. Vol.24, NO. 5, pp [9] Hong Wang and Bing LI. The cooperated MPPT Control of stand-alone PV power Generation System. Proceedings of the 8th world congress on intellignet Control and Automation, 2010, China. pp [10] H.Wang and Donglai Zhang. The stand alone PV generation System with Parallel Battery Chareger International conference on electrical and control enginerring, pp [11] Guo X. Yun, Chen Jinmei and Liu Qihiui. Real-Time and grid-connected control of PV power system. The international conference on advanced power systesm automation and protection pp [12] Himanshu Kumar and R.K.Tripathi. Simulation of variable incremental conductance method with direct control method using boost converter. Engineering and Systems 2012, India. pp

63 [13] M.Kaliamoorthy and V.Rajasekaran. A Novel MPPT Scheme for Solar Powered Boost Inverter using Evolutionary Programming. International conference on recent advancment in electrical, electronics and control engineering. 2011, India. pp Pages [14] Zhigang Liang and Rong Guo. High efficiency switched capacitor buckboost Converter for PV application. Applied Power Electronics Conference and Exposition (APEC), 2012 Twenty-Seventh Annual IEEE, 2012, USA. pp , [15] Jose C.U. Pena, Moacyr A. G., Guilherme de A. e Melo and Carlos A. A comparative study of mppt stratigies and a novel single-phase integerated buckboost inverter for small wind energy conversion systems. Power Electronics conference. 2011, Brazil. pp [16] Marcos Orellana, Stephane Petibon, Bruno Estibals and Corinne alonso. Four switch buck-boost converter for photovoltaic DC-DC power applications. IECON th Annual Conference on IEEE Industrial Electronics Society, 2010, France. pp [17] T.Mrabti, M.El Ouariachi, B.Tdhaf, Ka.Kassami and E. Chadli. Regulation of electric power of photovoltaic generators with DC-DC converter (buck type) and MPPT command. Multimedia Computing and Systems, ICMCS '09, pp [18] M. G. Villalva, J. R. Gazoli and E. Ruppert F. Analysis and simulation of the P&O MPPT algorithm using alinearized array model. Power electronics conference, 2009, Brazil. pp [19] Etrone, G., Spagnuolo, G. and Vitelli, M. Optimizing duty-cycle perturbation of P&O MPPT technique. Power electronics specialists conference, 2004, Italy. Vol.3, pp [20] Azadeh Safari and Saad Mekhilef. Simulation and hardware implementation of incremental conductance MPPT with direct control method using cuk converter. IEEE Transaction on industrial electronics Vol. 58 No 4. pp

64 [21] A.Safari and S. Mekhilef. Implementation of incremental conductance method with direct control. IEEE region 10 conference ICON, 2011, Malaysia. Pp [22] D.Menniti, A.Burgio, N.Sorrentino and A.Pinnarelli. An incremental conductance method with variable step size for MPPT design and implementation. 10 th international conference on electrical power and quality and utilization. 2009, Poland. pp 1 5. [23] Zhou Xuesong, Song Daichum, Ma Yoijie and Cheng Deshu. The simulation and design for MPPT of PV system based on Incremental conductance Method. WASE international conference on information engineering, pp [24] Jawad Ahmed. A fractional open circuit voltage based maximum power point tracker for photovoltaic arrays. 2 nd international conference on software technology and engineering pp V V [25] M.adly, H.el-sherif and M.Ibrahim. Maximum power point tracker for a PV cell using fuzzy agent adapted by the fractional open Circuit Voltage Technique. IEEE international conference on fuzzy systems, 2011, Taiwan. pp [26] Nelson Diaz, Adriana Luna and Oscar Duarte. Improved MPPT shortcircuit current method by a fuzzy shor-circuit current estimator. Energy converstion conference and exposition, 2011, Colombia. pp [27] Nelson Diaz, Johann Hernandez and Oscar Duarte. Fuzzy MPP method improved by a short circuit current estimator, Applied to a grid- connected PV system. IEEE 12th work shop on control and modeling for power electronics. 2010, Colombia. pp 1-6. [28] Subiyanto, A.Mohamed and M.A.Hannan. Maximum Power Point Tracking in Grid Connected PV System using A Novel Fuzzy Logic Controller. IEEE student conference on research and development, pp

65 [29] S.I. Sulaiman, T.K. Abdul Rahman, I.Musirin and S.Shaari. Optimizing Three-layer Neural Network Model for Grid-Connected Photovoltaic output prediction. Conference on innovative technologies in intelligent systems and industrial applications Malaysia. pp [30] Trishan Esram and Patrick L. Chapman. Comparison of Photovoltaic Array Maximum Power Point Tracking Techniques. IEEE Transactions on energy conversion, Vol.22, NO. 2, pp [31] Hannes Knopf,(1999). Analysis, Simulation and evaluation of maximum power point tracking (MPPT) methods for a solar powered vehicle. M.Sc. Portland state university. [32] M.I. Marei, E.F. El-Saadany and M.M.A. Salama. Flexible Distributed Generation: (FDG). IEEE Power Engineering Society Summer Meeting, Vol 1. pp [33] Haiyang Zang and Xiu Yang. Simulation of Two-level Photovoltaic Gridconnected System Based on Current Control of Hysteresis Band. Power and Energy Engineering Conference, 2011, China. pp 1-4 [34] KYOCERA KC200GT high efficiency photovoltaic module data sheet. [35] Mark Hudson Beale, Martin T. Hagan and Howard B. Demuth (2012). NNET 2012 Neural networks toolbox user s guide. 57

66 58

67 APPENDIX (A) A.1 ENERCON E-126 Data sheet 59

68 A.2 KYCOERA KC200GT Data sheet 60

69 61

70 APPENDIX (B) B.1 ANN training points 62

71 63

72 64