Optimizing direct normal irradiance prediction of BP neural network based on Logistic Sparrow search algorithm

Abstract With the wide application of photothermal power stations in systems, Direct Normal Irradiance(DNI) prediction is very important to improve economic benefits and utilization solar energy. The short-term wind by traditional Sparrow search algorithm (SSA) optimized BP(SSA-BP) neural network prone fall into local optimum, slow convergence rate, low accuracy. A DNI method based on Logistic (Logistic-SSA) BP(Logistic-SSA-BP) proposed. This paper takes a station Northwest China as research object. Firstly, Pearson correlation coefficient introduced analyze environmental data set with strong direct illumination radiation index input model, so avoid redundant affecting index. Secondly, SSA Logistic-SSA are used BP network. Finally, measured historical simulate models. simulation results show that model Logistic-SSA-BP has better accuracy than SSA-BP, more line production operation requirements station.