Building Energy Consumption Monitoring And Prediction

According to recent surveys, building energy consumption grows more and more rapidly, especially for public buildings, althrough their energy consmption gross accounts for a small part, the unit energy consmption of them is much higher than any other buildings, which provoked governments’ attention, furthermore, a series of policy and regulation were formulated. But without data support and scientific guide, it is hard to make energy using plan. Thus researchers brought many ways to solve this problem,however, these were still not qualified for requirements according to relevant authority. For example energy consumption information was acquisited by all kinds of ways without a standard method; historical information can’t be made good use and so on.Taking above as background, this research developed a set of building energy consumption monitoring system and established a building energy consumption prediction model. After studying previous prediction methods based on neural network, we pointed out the shortcomings of them. To make further study, we found out the existing optimization methods usually aimed at specific weakness of neural network without an overall improvement. However, short-term prediction of building energy consumption requires not only high accurate, it is also a kind of high real-time prediction. Therefore,neither traditional methods nor improved solutions can be qualified for it. To solve this problem, in this study we applied Levenberg-Marquardts(LM for short) algorithm to genetic algorithm neural network so that it would conbine the global search ability of genetic algorithm with local search ability of LM algorithm, and put forward GALM neural network learning algorithm. After being established, the prediction model based on GALM neural was tested to be enough accuracy and obviously efficient, which proves that the prediction model based on GALM neural networkis is applicable to make the short-term prediction of building energy consumption.In order to provide reliable data source for energy consumption prediction model, and improve the quality of energy consumption data, and realize intelligent management control combination, this research is to design and develop a set of energy consumption monitoring system.1. Considering the characteristics of installation and working environment, itpresent a design scheme of distributed monitoring and centralized management.2. Design energy consumption metering and management devices based on 51MCU;apply CS5460 A to acquisite instantaneous and effective values of energyconsumption information.3. Develop data acquisition program based on C language to acquisit both periodicaland real-time energy consumption data, and develop energy consumptionmanagement program based on PHP to make access for users to visit the dataresource through IE, or manage energy consumption by prediction model basedon GLAM.