Research on Multi-load Forecasting of Integrated Energy System Based on GRA-BP Neural Network

Aiming at the accuracy of regional comprehensive energy multi-load forecasting, a multi-load short-term forecasting model based on grey correlation analysis and improved BP neural network is proposed. The BP neural network prediction model has adaptive learning rate, elastically modifies the connection weight coefficient and improves the prediction accuracy. According to the typical climate characteristics in summer and winter, the correlation change law of cold, heat, electricity and gas load in an industrial park is dynamically simulated according to the actual calculation example. The actual results show that the improved BP neural network multivariate load short-term prediction model improves the prediction accuracy and has practical application prospects.