To ensure that the photovoltaic power generation system always works at its maximum power point, a method for photovoltaic maximum power point tracking by the neural network based on asymmetric basis is proposed, and its concrete implementation steps are given. Fuzzy factor membership functions are built according to the influences of photovoltaic power generation factors on the power generation efficiency, and the fuzzy weights of the influencing factors are calculated, with these weights infused into the building of the neural network based on fuzzy asymmetric basis. The network is trained by using methods of fixed basis width RBF, traditional RBF and the method proposed in this paper with four kinds of quantities of samples, and the comparison in terms of the network training time and the standard deviation indicates that the accuracy of the network with 180 samples is the highest, at least an order of magnitude higher than other that of methods. By determining the working parameters of the photovoltaic system in real time by using this neural network, the photovoltaic system can make the internal and external resistances completely match at every moment through the slide rheostat, thus ensuring that the system always works at the maximum power point.
