Application of RBF Neural Network in the Construction of Intelligent Predictive Model of Public Building Energy Consumption

Abstract

Energy saving and emission reduction are necessary means for our country to take the road of sustainable development. It requires the control of social energy usage and the active development of a low-carbon economy. For the sake of make a scientific prediction of structure energy usage, an intelligent forecasting model for public structure energy usage is constructed in accordance with RBF, and optimized by combining PSO algorithm and LM (Levenberg-Marquardt) algorithm. The results show that the SPO-LM-RBF forecasting model can get reasonable and accurate forecasting results of structure energy usage in both cooling season and heating season, the forecasting error is controlled below 2.1%, the average relative error is reduced by 2.24% and 1.33% compared with RBF neural network, and the daily maximum relative error is decreased by 4.75% and 3.76%, which is important to implement energy conservation and emission reduction of public structures.