Short Term Power Load Forecasting Model Based on XGBoost-BiLSTM

Short-term power load forecasting is able to effectively help power companies to map out electricity generation schemes, allocate grid resources, provide system security analysis and facilitate power market transactions, which can help to ensure a stable and reliable supply of electricity, and has become more and more important in the operation and management of power supply enterprises. In order to upgrade the precision and economy of power load forecasting and enhance the intelligent level of load forecasting, this article establishes a short-term power load forecasting model on the basis of XGBoost-BiLSTM(eXtreme Gradient Boosting- Bi-directional Long Short Term Memory). First, the load data is preprocessed to remove the dirty data that affects the forecasting results. Then the load data is input into XGBoost and BiLSTM models for feature learning. Finally, Stacking is used to fuse the two training models. The research results show that XGBoost-BiLSTM short-term power load forecasting model can improve the precision of load forecasting and provide effective help for power load forecasting.