An integration of enhanced wind power interval forecasting into reactive power dispatching

The variability and unpredictability of the wind power has inevitably triggered a number of issues in the operation of the electrical power system such as voltage stability, power flow distribution and economic efficiency which draw researchers´ attentions on the reactive power dispatching of wind farms. A traditional wind power forecasting is one of an essential way alleviating the above issues due to its contribution on deterministic dispatching. However, because of the unreliably predictive nature of wind power, the integration of wind power requires both the improvement of forecasting accuracy and the consideration of wind power forecasting uncertainty into reactive power dispatching. In this paper, an assimilation of an enhanced wind power interval forecasting system and reactive power dispatching has been proposed. The improvement of wind power forecasting is addressed from NWP amendment model and the interval forecasting model using least square method and relevance vector machine respectively. Furthermore, the forecasting and its uncertainty results from the interval forecasting system are integrated into the reactive power dispatching under a certain confidence level. The dispatching model is established considering the lifetime of capacitor and power loss of within wind farm. To take a wind farm in northwest of China as validation, the results show that the proposed interval forecasting system outperforms GA-BP model and SVM model, while the reactive power dispatching model combining the forecasting and its uncertainty information with has advantages over the traditional counterparts in terms of risk resistance, wind farm reliability and economic efficiency.