Bus Reactive Load Prediction Based on Random Forests

Abstract

The change of reactive load is complex and the power factor changes at any time. The establishment of accurate reactive load forecasting model can not only improve the reactive power management level of the grid, but also play an important role in intelligent dispatching, automatic voltage control and economic operation of the system. At present, reactive power optimization of online power grid mostly uses real-time data obtained by SCADA system, which belongs to post-correction control, and the control strategy has poor timeliness, so it is difficult to deal with the current situation of power grid operation with increasing uncertainty of both source and charge. In this paper, the characteristics of reactive load are analyzed, and a method of bus reactive load prediction based on Random Forest (RF) is proposed, which takes online reactive load obtained by SCADA as input and output data, uses historical data to train the RF reactive power prediction model to obtain the reactive power load the next day. The experimental results show that this method can accurately predict the reactive load of bus and makes a contribution to the initiative and refinement of power grid regulation.