Short-term Power Prediction Of Photovoltaic Power Station Based On PSOEM Algorithm And Neural Network

The depletion of fossil energy and the pressure of environmental protection have forced mankind to develop and utilize new clean energy sources,Photovoltaic power generation,as a new type of clean energy,has been widely used and developed rapidly in recent years.The power of photovoltaic power generation has random fluctuations.The accurate prediction of the output power of photovoltaic power station can help the power dispatching department to arrange and control the operation mode reasonably,and provide the guarantee for the economy and reliable operation of the power system.This paper first introduces the principle of photovoltaic cell generation,the composition of grid-connected photovoltaic system,and then makes a careful analysis of the photovoltaic power prediction and the main factors affecting the photovoltaic power prediction;next,the BP neural network in artificial neural network is studied and analyzed,and the power prediction model based on BP neural network is determined,including the number of hidden layers and the number of input and output nodes,as well as the selection of incentive function.Aiming at the situation that the historical data of output power of photovoltaic power station have the potential for local vacancies,out-of-scope and abnormal fluctuation,a method of data detection and repairing the detected abnormal data is proposed to improve the accuracy of power prediction of photovoltaic power station.On this basis,two swarm intelligence algorithms,Particle Swarm Optimization algorithm(PSO)and Particle Swarm Optimization with Extended Memory algorithm(PSOEM),are used to optimize the initial and threshold values of BP neural network for the problems of easy falling into local extremum and slow convergence.The basic principle and implementation steps of PSO and PSOEM are introduced,and the output power prediction model of photovoltaic power station with PS O-BP neural network and PSOEM-BP neural network is establishedFinally,according to the historical data of photovoltaic power generation from February 1 to June 30 in a photovoltaic power station in Qinghai Province,the power prediction of photovoltaic power generation system is carried out by using the three models proposed.the error comparison results show that the accuracy of power prediction of PSOEM-BP neural network is obviously higher than that of PSO-BP neural network and it is of certain theoretical and practical value to use PSOEM optimized BP neural network model to predict photovoltaic power.