The Research Of MPPT Of PV Power Generation Based On Neural Network

With the rapid development of modern economy, problems of energy shortage and environmental pollution are becoming more and more serious. Solar energy resource has been paid attention to the whole world because it is abundant, clean and safe. So all the countries have increased the investment of photovoltaic research, but high investment and low efficiency of photovoltaic industry can not be ignored. In order to solve this problem, maximum power point tracking will become one of the hot researches to photovoltaic. This paper proposes a method of maximum power point tracking based on neural network from the angle of improving the photoelectric conversion efficiency. Improving the utilization of photovoltaic cells is of great significance for better development of the optoelectronics industry.At first, this article introduces the background and the significance of photovoltaic research, and introduces the research status of photoelectric at home and abroad. Then the paper gives a detailed introduction of the system structure and classification, introduces three kinds of photovoltaic cell model, and every model is built up. Using these models study light intensity and temperature on PV cell characteristic influence. This paper introduces the MPPT working principle and implementation methods of photovoltaic system, and analyzes the conventional MPPT control methods, at the same time, lists the advantages and disadvantages of various algorithms. The above-described methods have some problems, like slow response, shocking at the maximum power point and bad accuracy. So this paper proposes that using neural network achieves PV MPPT, and gives a detailed design, and completes modeling and simulation by MATLAB/SIMULINK. The simulation result shows that maximum power point can be tracked exactly and quickly by neural network. In order to improve neural network, the networks are in series. Compared to the previous independent network, the simulation result shows that the generalization performance of the network which in series has been improved and the fitting effect is better. Thus it is completely feasible to be applied the solar energy power generation system.