Research on Photovoltaic Power Prediction Method for Power Grid Safety

Abstract

When integrating large-scale photovoltaic systems with the power grid, variability and intermittency of photovoltaic power may potentially endanger the secure and stable operation of the power system as well as its scheduling management. So a new photovoltaic power prediction method using logistic chaotic mapping (LCM) improving atomic search optimization algorithm (ASO) to optimize back propagation neural network (LCM-ASO-BPNN) is proposed to solve this problem. The ASO algorithm is used to solve the defect that BPNN is likely to be trapped in a local optimum, and the initial population of the ASO algorithm is optimized by introducing logistic chaotic mapping, subsequently, the model's predictive accuracy is greatly enhanced. The experimental results demonstrate a significant improvement in the prediction accuracy of the proposed model when compared with the traditional prediction model.