Application of multi-expert information fusion prediction model to forecast station power consumption

Station power consumption forecasting can provide a strong basis for the pre-diagnosis of substation equipment faults, which is of great significance to improve the safety and reliability of power system. In this paper, support vector machine, gray theory and BP neural network prediction algorithm are used to calculate their respective weight coefficients according to the principle of the smallest sum of squared prediction errors. Weighted establishment of multi-expert information fusion prediction model is calculated to accurately predict the power consumption of substations. The multi-expert information fusion forecasting model integrates the advantages of three single-item prediction methods, and accurately forecasts the station power consumption of a 500kV substation in Guangdong Province. The average relative error is 5.8%, which is less than the average relative error of the prediction results of each individual module. The analysis of prediction examples proves the reliability and effectiveness of the multi-expert information fusion prediction model.