Research on Wind Power Prediction Method Based on Convolutional Neural Network and Genetic Algorithm

Abstract

With the vigorous development of renewable energy, large-scale wind power is connected to the power grid. Compared with conventional energy sources, wind power generation has the stochastic and intermittent characteristics, which brings great challenges to the safe and stable operation of power grids. Forecasting is essential for the rational, safe and effective use of wind power. Thus, the ability to predict wind power in a timely and accurate manner is particularly significant.The short-term prediction error of wind power is mainly caused by inherent stochastic factors and extrinsic stochastic factors. The inherent stochastic factor refers to the defect or imperfection of the prediction system itself. The external stochastic factor refers to the imperfect data input by the system or the error of the input data. Artificial neural network (ANN) algorithms are supposed to be with better performance to solve such problems. This paper reviews the development of wind power generation and the research status of wind power prediction technology. On this basis, a wind power prediction model of convolutional neural network (CNN) and genetic algorithm (GA) is constructed to predict the short-term wind power. The learning accuracy, the number of hidden nodes, and the training function are discussed. The selection of the optimal neural network and GA prediction model are presented. The simulation results show that the prediction accuracy and stability of the proposed prediction system are improved, compared with the BP neural network and CNN.