The Forecasting Model of Reactive Power Based on SVM

With the development of distributed power generation and microgrids, the demand for fine-grained control of the power grid is becoming more and more urgent. Power grid reactive power has the characteristics of complex changes and fluctuating power factor. However, the current forecast of power grid compliance only focuses on active prediction. This paper proposes a method of grid reactive power prediction based on support vector machines. This solution uses EMS to obtain load data online, and uses historical data to train the SVM reactive power prediction model to obtain the reactive power load the next day. By predicting the reactive power of a power station in Shanxi Province, the accuracy of the model prediction was successfully verified. The invention fills in the blank of reactive power prediction and contributes to the initiative and refinement of grid regulation.