Short-term Generation Power Forecast Of Photovoltaic Power Station Based On GA-BP Neural Network

With the rapid development of society,users’ demand for electricity continues to grow,and clean renewable energy solar power generation has attracted more and more attention.However,the disadvantages of solar power generation are also very obvious,such as the volatility and intermittency of photovoltaic power generation output,which will cause a huge impact on the main network and is not conducive to maintaining the stability of power grid operation.In this situation,people’s demand for power generation prediction of photovoltaic power stations arises at the historic moment.In this paper,the genetic algorithm optimized BP neural network(ga-bp neural network)is used as the final prediction method.First,the power generation data of a 110kV photovoltaic power station for a whole year was obtained.Combined with the established solar cell model,the factors(solar irradiance,weather type,seasonal type,temperature and humidity)affecting the output power of the photovoltaic power station were studied.Then,the output power prediction model of photovoltaic power station based on BP neural network is established,and the specific network topology,learning process and number of neurons in each layer of the network are described in detail.In the process of using BP prediction model to predict,it is found that BP neural network has disadvantages of multiple iterations and long convergence time.In order to overcome these disadvantages,this paper proposes to use GA genetic algorithm to optimize the weight and threshold of BP neural network,and establishes ga-bp prediction model.Finally,the two prediction models are respectively used to predict the power generation of a photovoltaic power station.According to the error analysis of the prediction results of the above two prediction models,it is determined that both models can predict the output power of photovoltaic power stations,while the ga-bp prediction model has better stability and higher prediction accuracy.This result proves that the proposed method of short-term prediction of photovoltaic power station output power by ga-bp neural network is correct and feasible.