Non-intrusive Load Decomposition Considering Electrical Features and Non-electrical Features in the Combined Optimization Method

Abstract

Non-intrusive load monitoring technology is of great significance for carrying out energy conservation and emission reduction work, obtaining the electricity consumption information of appliances and building smart grids. In the current load decomposition method, the combination optimization method only considers the traditional electrical features, which leads to low decomposition accuracy. Therefore, it is necessary to introduce new features to improve the load decomposition accuracy. In this paper, based on the traditional power features, the state transition rule of electrical appliances and user habits are considered, and non-electrical features are introduced into the load decomposition method, which requires less prior information to achieve load decomposition. Finally, the feasibility and effectiveness of the proposed method are verified based on the common data set AMPds.