Research And Implementation Of A Building Energy Consumption Analysis Model Based On BP Neural Network

Currently, with the vigorous push of China’s urbanization process and the increase of urban population, the total energy consumption of buildings grows sustainability, and the growth rate has an upward tendency. The building energy consumption accounts for 32% of the total energy consumption of the community, and public building energy consumption accounts for 22% of the total building energy consumption[1], it has become the focus areas of energy conservation. Faced with such serious situation of the building energy consumption, the traditional energy consumption simulation analysis method due to its great limitation and low accuracy, already cannot satisfy the existing user requirements. In order to better help people understand the current energy consumption situation of buildings accurately and effectively, determine the appropriate improvement measures to save energy, and provide decision support for the daily energy saving management, in this paper, a building energy consumption analysis model based on BP neural network is proposed by researching and analyzing the building energy consumption indicator data and its influencing factors, to locate the unreasonable links in the process of building energy consumption.First of all, this paper presents overview of research in field of the domestic and foreign building energy consumption analysis model, and compares the different emphasis of them.In the following part, by looking at the existing normative documents, literature material, and energy saving buildings case reports, consulting experts for experience, and analyzing the building energy consumption indicator data and analysis methods, the analysis data of the model is determined, and the corresponding relationships between the phenomenon of building energy consumption indicator and unreasonable links in the process of building energy consumption are summarized, providing the basis for the implementation of building energy consumption analysis model.Moreover, this paper proposes a analysis model which can analyze the existing unreasonable links in the process of building energy consumption automatically based on BP neural network. In the designing process, on the one hand, aiming at the disadvantages of the BP neural network, training more and the slow convergence speed, put forward to two improvement strategies base on the characteristics of the building energy consumption indicator data, optimize the initial weights and the factor of learning rate; On the other hand, for the problem of low analytical accuracy by using ordinary normalization, the binarization idea of image processing is employed to preprocess the model input data, effectively improve the analytical accuracy of the model.Finally, implement the building energy consumption analysis model, and test it on the existing buildings, to verify the accuracy and practicality of this model, at the same time, it is proved to be effectively helping people understand the current energy consumption situation of buildings, and providing decision support for the daily energy saving management.