Method Research And System Development Of Energy Analysis For Public Buildings Base On Data Mining

Currently, the building energy consumption accounts for32%of the total energyconsumption of the community, and public building energy consumption accounts for22%of the total building energy consumption, it has become the focus areas of energyconservation. With the popularity of energy subentry measure in many cities, moreand more energy consumption data was collected and stored. It has become a hotresearch topic to analyze energy consumption objectively and accurately and providedecision support to energy conservation. The energy consumption data contains therich information which is useful to save energy, however, large amounts of data hasalso brought a "data disaster", which is difficult to find potentially useful knowledgequickly and efficiently. There are some data mining methods have been proposed, butthese methods did not consider operating rules of buildings, and most of them studyon the specific building for static analysis, which result in deficiencies in theversatility and real-time.In this paper, an energy consumption analysis model based on data mining wasproposed, including energy consumption monitoring model, prediction model andevaluation model. The energy consumption monitoring model first identify the energyconsumption patterns by clustering analysis of historical energy consumption data, getthe decision tree of energy consumption pattern by classifying the energyconsumption data, matches the real-time energy consumption data with the energyconsumption patterns, and then make outlier analysis with historical data of the samepattern to determine whether the current energy consumption is abnormal. The energyconsumption prediction model first extracts the energy data of the same pattern byrecognize the energy pattern of the time to be predicted, then divide it according tocertain rules and build the RBF network to predict energy consumption value of thetarget time. The energy consumption evaluation model first extract attribute for thesample building energy consumption data set, then build RBF neural network andtrain it use these data to predict energy consumption value of the target building, anddetermine the energy situation by compare the predicted and actual value.In order to apply the model to practice, the energy consumption analysis systemof public building was developed with Visual C++2008. This system can help theadministrator to monitor, predict and evaluate the energy consumption easily, without complex operations and the relevant professional knowledge.The validity of the model was verified by apply this model to the summer energyconsumption data of a comprehensive building and the U.S. commercial buildingenergy consumption statistics database (CBECS). Apply the energy consumptionanalysis system to analyze the actual energy consumption of a building, result showsthat the system can work for energy monitoring, prediction and evaluation, and has ahigh operating efficiency.The energy consumption analysis model presented in this paper is versatile andreal-time. It can be apply to the analysis of the actual energy consumption effectively,which is helpful to building energy conservation.