Wind speed prediction research based on time series model with residual correction

With the sharp reduction of global fossil energy and the deterioration of the environment, governments focus on clean renewable energy. Wind energy as a green environment-friendly energy is subject to widespread concern, however, the high degree of randomness and volatility of wind restrict the healthy development of wind power industry. Increasing the short-term wind speed prediction accuracy is the key to guarantee the stable operation of the power system and effectively adjust the scheduling plan. The traditional prediction algorithms are mostly limited to the improvement of the algorithm itself, ignoring the analysis and utilization of the prediction residuals. Therefore, the paper uses the time series model of BP neural network residual correction to predict wind speed. The result indicates that the time series model with residual correction can reflect the actual law and improve the predicting accuracy effectively. The research work will be helpful to the reasonable arrangement of dispatching operation plan, reduce the cost of electricity operation, improve the market competitiveness of wind power.