Wind Speed Prediction of IPSO-BP Neural Network Based on Lorenz Disturbance

Abstract

Accurate wind power prediction provides significant guarantee for power grid dispatching, and wind speed prediction, as the basic link of wind power forecasting, has crucial theoretical research significance and practical application value. In this paper, we present the wind speed prediction of IPSOBP neural network based on Lorenz disturbance. At first, the data is processed by principal component analysis (PCA) to select the key factors affecting wind speed, which can effectively reduce the complexity of model. Then, the improved particle swarm optimization (IPSO) algorithm is used to globally optimize the weights and thresholds of BP neural network, and overcome the problem of local minimum value. The initial prediction results can be obtained by the IPSO-BPNN model. Finally, Lorenz system is introduced to correct the initial prediction value and improve forecasting accuracy. According to the wind farm data of Spain and Chang Ma in China, we take an empirical research to analyze the optimization effect of IPSO algorithm and the promotion effect of Lorenz system on the precision of preliminary forecasting. The results are as follows: 1) IPSO algorithm accelerates the convergence rate of weights and thresholds of BP neural network and 2) Lorenz disturbance system obviously weakens the random volatility of wind speed, effectively modifies its preliminary prediction results, and upgrades its prediction accuracy.