RBF Natural Network Based On Genetic Algorithm Used In Maximum Power Point Tracking Of Photovoltaic System

With the increasing consumption of traditional fuel and rising demand for energy, renewable energy draws more and more attention. As a kind of safty and renewable energy, solar energy has a widely application prospect that has got rapid development in our country. But the low efficiency and high cost factors have seriously restricted the promotion and use of the photovoltaic power generation system at present, so how to improve the efficiency and reduce the cost of power generation becomes a key problem to research. As the maximum power point tracking(MPPT) is an important way to improve the conversion efficiency, this paper presented a new control method which is based on genetic algorithm to optimize the RBF neural network by using the neural networks’ fitting and prediction ability and genetic algorithm optimization outstanding characteristics, to realized the MPPT in photovoltaic power generation system.Firstly, this article is introduced the current photovoltaic background and the trend of photovoltaic power generation at home and abroad, then the significance and methods of MPPT are illustrated. Setting up the photovoltaic power generation model by MATLAB, then the U-I and P-V curve will be obtained by the simulation, and it is researched that the sunshine intensity and temperature are two main variables to influence the output power of maximum power point.Secondly, the principle of the MPPT is expounded. Then the advantages and disadvantages of traditional tracking method and its improved method are illustrated, especially, the artificial neural network control method are emphatically introduced in details. Through the comparative analysis of different neural networks, radial basis function(RBF) neural network is chosen to predict the output value of maximum power point of photovoltaic power generation.Thirdly, for the defects of RBF network structure, this paper uses genetic algorithm to optimize the centers, widths, and the weights which is the connection of hidden layer and output layer. The Genetic Algorithm, a global optimization algorithm simulating organic evolution, is used to optimize the parameters of the RBF network. The coding method put all the parameters in one chromosome optimize them synchronously and strength the cooperation between the hidden layer and the output layer. In addition, the genetic mechanism of the genetic algorithm itself has made the corresponding improvement, so as to further perfect the genetic operation.Finally, by using RBF neural network that optimized by the genetic algorithm and the traditional RBF neural network to forecast the actual maximum power point of photovoltaic power generation system, Results show that the optimized RBF neural network to achieve the target of error before training number is significantly reduced, average error was reduced by 3.7%, so the optimized genetic algorithm of RBF neural network not only improves the speed of prediction, but also the prediction precision, which can better control the photovoltaic maximum power point tracking.