Research on Short-Term Power Load Forecasting Model Under Integrated Energy Optimal Scheduling

This research focuses on improving the level of intelligent control of power systems, with special emphasis on accurate prediction of short-term power loads to meet the growing power demand and improve the quality of power supply. In the field of short-term power load forecasting, many factors affect the accuracy, so the core problem of this study is how to improve the accuracy of load forecasting. Two improved short-term load forecasting methods are proposed. First, the error compensation kernel random weight neural network is used to establish an input index system including seasonality, weather change, industrial production and other factors to improve the forecast accuracy. Second, the method based on multivariate chaotic time series and weighted kernel random weight neural network aims to reduce the prediction error caused by chaotic dynamics and sample weights, and better capture the complex dynamic changes of power loads. A highly accurate power load forecasting model with low levels of RMSE and MAE was obtained. Compared with the existing methods, these two improved methods have obvious advantages in accuracy, generalization performance and training efficiency, and provide a reliable tool for intelligent control of power system and promote the sustainable development of power system.