A Novel Short-term Wind Speed Prediction Method Based On Rolling Forecasting And Its Error Analysis

In today’s power system industry, the microgrid can promote renewable energy and distributed generation network, it is conducive to the development of renewable energy in China. Power system management at the edge of a large number of distributed power grid may cause uncontrollable power system, insecurity and instability, thus affecting the operation of the power grid and power market transactions, so distributed generation faces many technical hurdles and challenges. For wind power technology, it is one of the microgrid key research issues. For these reasons, short-term wind speed prediction model is important.For a microgrid with wind turbines, how accurately and effectively for the next 24 hours and fan output forecast microgrid rational and efficient allocation of energy will be the key to become microgrid energy storage optimized configuration. Therefore, this article on this subject carried out corresponding research, the study of wind power prediction error model, combined with rolling forecast model, constitutes a novel method of rolling forecasting and interval error correction model portfolio. Meanwhile, according to the prediction error probability distribution produced by microgrid energy distribution of prediction, so as to better micro-grid energy storage configurations accurate planning. The main contents and innovations of this paper are summarized as follows:(1)Short-term wind speed forecasting models are summarized. The models divided into single forecasting model and combined forecasting model. In each class predictive models, using different forecasting methods to predict the results of a detailed quantitative analysis and comparison.(2)The error model of wind speed forecast method is studied. And a novel combined forecasting approach based on interval error correction and rolling forecasting is presented. By the proposed method,24-hoour-ahead wind speed forecasting is divided into five rolling steps over different corresponding time horizons in advance. The error interval correction models are used in every single step. With the number of steps by scrolling, every step of the error has also been corrected and gradually improved, and the final predicted results over five different time horizons maybe brought to the system operator step by step for scheduling strategies decision making.(3)Because of uncertainty for wind power, the microgrid wind turbine output prediction error probability distribution models are studied. And under the conditions of a given degree of confidence, the confidence interval of prediction error is calculated.(4)The history data in three years are used for the proposed forecasting methods. The forecasting results of hourly forecast, monthly forecast and seasonal forecast are presented, respectively. By the performance, it can be seen that the proposed method show a significant improvement in prediction error. And it also proves the versatility of the proposed method. Meanwhile, probability distribution of wind power forecasting error is used to quantitative analysis the distribution of energy storage in a microgrid. Finally the configurations of microgrid energy storage are optimized.