Study On Energy Consumption Prediction And Information Management System Based On Deep Learning For Large-scale Public Building

Large-scale public building only occupy about 4% of the total construction area of China.However,its energy consumption is as high as 22% of the total energy consumption of the building,which represent the low energy usage efficiency.Hence,research on conservation and reduction of energy for large-scale public construction is more important.Accurate prediction of building energy consumption and refine energy consumption supervision not only achieve scientific planning of energy usage from the building energy supply side,but also provide important data for the fine management of energy usage from the building energy demand side,which has great meaning of solving the low energy usage of large-scale public building.In response to the above problems,this paper takes a large-scale public construction in Shaanxi Province as the research object,and the main research are as follows:(1)Firstly,the reasons for affecting the energy consumption of large public buildings are analyzed,determine the content of large-scale public construction energy consumption measurement according to relevant policy guidelines and propose renewable energy measurement methods.Then,design the Internet of Things mode of the large-scale public building energy consumption supervision system and explaine the functions of each part of the system.Finally,the ZigBee-based acquisition terminal and gateway node according to the requirements of the building energy supervision system is demonstrated.(2)Aiming at the massive mass,complex nonlinearity and time series of large-scale public building energy consumption data,using deep learning powerful nonlinear mapping ability and feature extraction ability to establish a large-scale public building energy consumption prediction based on deep learning The model is validated by an example.The experimental results show that the MRE index of the proposed model is 4.3784% lower than the linear regression model,4.1913% lower than the BP neural network model,and 0.1167% lower than the support vector regression model.It is confirmed that it has higher accuracy in building energy consumption prediction.(3)Based on the analysis of the large-scale public construction energy consumption Big Data characteristics,the cloud computing data center of the large-scale public construction energy consumption supervision system including the virtualization infrastructure layer is researched,the cloud computing basic platform layer and the cloud computing information service layer.Thus,the processing,storage and management of large-scale public construction energy consumption big data is realized.Using SQL Server 2008 software to build a large public building energy consumption database.Based on the B/S architecture,the energy information management system is designed and developed in conjunction with the Java language.The system can satisfy the requirements of real-time supervision of energy consumption.This study will help to understand the energy consumption of large-scale public buildings in real time and grasp the trend of energy usage.Apart from this,it will provide data and theoretical support for the development of building energy conservation work.