A Framework of Day-Ahead Wind Supply Power Forecasting by Risk Scenario Perception

Abstract

Wind power forecasting (WPF) systems are essential to maintain the safe and stable operation of the power system in case of large-scale grid-connected wind farms. However, the current forecasting has the problem of disunity between statistical value and application value, that is, it only pays attention to its forecasting accuracy and ignores the risks caused by it in the power system. In order to solve the above problems, this study proposes a framework of wind supply power forecasting (WSPF) for wind farm cluster, which takes into account the risk scenario perception. First of all, aiming at the predicted risk phenomenon in WPF, TimesNet combined with the fluctuation information of Numerical Weather Prediction (NWP) wind speed is used to identify the corresponding risk scenarios. Secondly, the effective consumption area and power supply risk area evaluation index, as well as the accuracy of WSPF are defined, and the optimal forecasting curve correction scheme is fitted according to the index. Thirdly, taking into account the correction scheme and identification results, a variety of predictors are used to verify the WSPF according to the above framework. Finally, the proposed method is applied to a wind farm cluster in Inner Mongolia Autonomous region of China, the average accuracy of WSPF has increased by 37%, which verifies the effectiveness and universality of this method.