Research On MPPT Control For Photovoltaic Power System

With the economic growing and the rapidly growth of world population, the demand for energy of the society is growing and the traditional fossil energy is not suitable to sustainable development. The traditional fossil energy reserves are limited, and the use of the traditional fossil energy cause serious environmental pollution, so it is urgent to develop new energy. Solar energy is a non polluting green, sustainable energy so it is necessary to research the maximum power point tracking (MPPT) of photovoltaic cells to improve the power efficiency of photovoltaic power system.Establishing the photovoltaic battery equivalent circuit and the equivalent output mathematical formula and a Matlab simulation model of photovoltaic cells to study maximum power point tracking of photovoltaic power generation system, Analysis of the two factors that affect the output characteristics of photovoltaic cells light intensity and cell temperature, and studying the P-V and I-V characteristics of photovoltaic cells with the changing two factors.Study the PV cell characteristics and introducing the control methods of MPPT of photovoltaic power system. To study the perturbation and observation method and incremental conductance method which based on self optimizing perturbation control method, establishing circuit simulation model of the two methods, and analysis these two method. Solve the misjudgment and step set issues, and simulate the improved methods on Matlab. The results show that they can effectively increase the output power and improve power generation of photovoltaic cells.Chose support vector regression (SVR) to predict maximum power point of PV. Establishing the SVR forecasting model, and on this basis compiling programphotovoltaic MPPT controller based on SVR, realizing the establishment of the application based on MPPT controller SVR in photovoltaic power generation system specific circuit, results verify the feasibility and effectiveness of photovoltaic MPPT controller based on SVR.Compared with forecasting model based on BP neural networks, the forecasting model based on SVR control method prediction accuracy is better, so the SVR forecasting model is more suitable for photovoltaic MPPT controller. Because SVR prediction model parameters has great impact on the predicted performance so use the grid search algorithm (GSA), genetic algorithm (GA) and particle swarm optimization (PSO) to optimize model parameters and comparison performances of these three algorithms. The results show that the GSA-SVR method is better than the others, but it needs to spend a lot of time; PSO-SVR method prediction accuracy is slightly lower than the GSA-SVR, but higher than that of GA-SVR, PSO-SVR prediction model optimization is simple, short time, high precision, so it is more suitable for photovoltaic MPPT controller optimization PSO algorithm used to establish the SVR prediction model.