Application Research On Building Energy Consumption Data Analysis Based On Data Mining Technology

As the acceleration of urbanization and the improvement of people's living standard in China,the proportion of building energy consumption to the total energy consumption of society is about 30% so far.Therefore,the reduction of building energy consumption is important to build a conservation-minded society.More and more building energy consumption data are collected and stored in the database of building energy consumption monitoring platform,but traditional monitoring methods of building energy consumption are difficult to find the features of energy consumption data and the potential relationships among energy consumption data effectively,and is lack of accuracy of the identification of abnormal energy consumption data.Accurate results cannot be analyzed by single intelligent analysis method of building energy consumption.Based on the analysis of the characteristics of building energy consumption and the analysis methods of building energy consumption,an integration intelligent technique of building energy consumption analysis based on many kinds of data mining algorithms(IIT)was proposed.The valuable knowledge concealed in building energy consumption data can be explored by the technique with the integration intelligent of suitable data mining algorithms which were optimized from the following respects: classification,outlier analysis,association analysis and prediction.There were four algorithms in the IIT of building energy consumption analysis.Clustering algorithm can be used to build the patterns of building energy consumption.Outlier analysis algorithm can be used to identify the abnormal energy consumption data.Association analysis algorithm can be used to find the related factors affecting energy consumption.Prediction algorithm can be used to predict energy consumption.IIT was used to compare with LOF outlier algorithm and BP neural network algorithm by the comparison experiments.The comparison experimental results were listed as follow: the accuracy rate of IIT is 62.9% higher than LOF outlier algorithm in the comparison experiment of abnormal data recognition,and the accuracy rate of IIT is 13.85% higher than BP neural network algorithm in the comparison experiment of energy consumption prediction.The experiments results show that the accuracy of IIT is higher than the single intelligent data analysis method in the analyses of historical building energy consumption data.In order to certificate the effectiveness of IIT of building energy consumption analysis,an intelligent system of building energy consumption analysis was developed rely on a monitoring platform of energy consumption.The system consists of four modules for building energy consumption analysis including: the analysis module of pattern classification,the analysis module of outliers,the analysis module of association and the analysis module of prediction.The feasibility of the system were certificated by the system running results based on the building energy consumption data set of the national renewable energy laboratory in USA.The test result shows the effectiveness of IIT based on data mining algorithms.IIT can be applied to the intelligent analysis of various types of buildings to support the decision making of building energy saving.