A nonintrusive power load monitoring method using coupled allocation mechanism

Abstract

 For non-intrusive power load monitoring problem, the key trouble is that there contains complex multiple types of appliances in a power load environment. In this study, two key suppositions are firstly introduced: (1) the signal characteristics should keep stable for each load appliance with a same running state in continuous times; (2) there is at most one running state change at an enough small monitoring period. Then, we consider that a probabilistic label value for each possible load can be evaluated according to a probabilistic clustering principle. Moreover, a coupled allocation mechanism on mixed probabilistic labels is introduced, in which an iterative filtering strategy is designed to estimate the optimal state combination of different loads. By performing professional load scenario simulations, the algorithm performance is effectively examined. The corresponding results indicate that better comprehensive performance can be obtained by the proposed method compared to the latest hidden Markov model and fuzzy clustering method.