A novel model for wind power forecasting based on Markov residual correction

An accurate wind power forecasting model has great significance in wind farm operation and electric power system dispatching and operation. An auto regressive integrated moving average (ARIMA) time series model with Markov residual correction is proposed to forecast the wind power in this paper. After establishing ARIMA model, random residual sequence with Markov property can be proved through chi-square statistics. The residual correction model based on Markov chain is then established. The prediction results of wind power of two wind turbines and the wind farm are achieved. The results with the assessment of accuracy rate and qualification rate show that the proposed model has excellent performances and precision. Compared with time series and artificial neural network model, the accuracy is improved by 6-10%, and qualification rate is improved by 2-7%. The proposed method implements more simply than some combined models, which has better practical value.