Graph representation learning-based residential electricity behavior identification and energy management

Abstract It is important to achieve an efficient home energy management system (HEMS) because of its role in promoting saving and emission reduction for end-users. Two critical issues HEMS are identification user behavior strategy. However, current methods usually assume perfect knowledge or ignore the strong correlations usage habits with different applications. This can lead insufficient description suboptimal To address these gaps, this paper proposes non-intrusive load monitoring (NILM) assisted graph reinforcement learning (GRL) intelligent decision making. First, a correlation incorporating NILM introduced represent consumption users multi-label classification model used monitor loads. Thus, state transition be achieved. Second, based on online updating graph, GRL proposed extract information contained graph. reliable strategy under uncertainty environment available. Finally, experimental results several datasets verify effectiveness model.