Building Energy Consumption Prediction of Housing Industry in China Based on Hybrid Models

Building energy consumption is a vital part of the total energy consumption in China, it is meaningful to predict the building energy consumption exactly as it is useful in the effective implementation of energy policies and is propitious for further expansion of the housing industry. In this paper, based on the factor analysis theory to reduce the dimension of the building energy consumption index, hybrid models of BP neural network and Least Squares Support Vector Machines are constructed respectively to predict the building energy consumption. Relevant data is collected from National Bureau of Statistics of China (1981 similar to 2009). Data analysis shows the proposed models, especially based on LS-SVMs, have more steady performance and higher accuracy.