Probabilistic graphical model based residential energy behavioral analysis on hybrid computing platform

The accurate customers energy behavioral can be analyzed from large amount of data acquired by smart meters. Practically, an ideal analysis system should meet two kinds of data demand: fast with low accuracy and slow with high accuracy. In this paper, a hybrid computing platform to process large amount of real-time data is proposed to analyze customer's energy behavioral. The structure and function of the platform is introduced. A graphical model based power consumption characteristics analysis algorithm is proposed. The implementation of the algorithm on hybrid computing platform is introduced. With energy usage data, four characteristics of resident energy consumption models are clustered. The experimental results show that the proposed algorithm can improve the efficiency of mass data clustering analysis and also proves the feasibility of the model.