Research on wind power Prediction based on BP neural Network

Abstract

Accurate prediction of wind power is beneficial to relieve the peak load of power grid and improve the capacity of power grid to accept wind power. Aiming at the problem of low precision of wind power prediction, this paper proposes an improved BP neural network prediction method, namely, an iterative genetic optimization BP neural network power prediction model with the power one hour in advance and other influencing factors as input. In this paper, wind speed, wind direction, temperature, humidity and air pressure are selected as the input data of the model, as well as the wind power of the previous period which is highly correlated with wind power. Since it is very unlikely that the values of adjacent moments will change before, the power of the first 15 minutes, the last hour and the previous day are selected in this paper. Finally, this paper analyzes and compares the above three situations, and the experiment shows that the model can effectively meet the relevant prediction demand of the power system for the actual short-term power of the wind farm.