Shengchun Yang’s research while affiliated with China Electric Power Research Institute and other places

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

Abstract To reduce the total power generation cost and improve the frequency stability of an island microgrid integrating renewable energy generation sources, a data‐driven cooperative load frequency control (DC‐LFC) method is proposed for solving the coordination control problem occurring between the controller and power distributor of the system. A novel algorithm, termed the effective exploration‐distributed multiagent twin‐delayed deep deterministic policy gradient (EED‐MATD3) algorithm, is further proposed, the design of which is structured based on the concepts of imitation learning, ensemble learning, and curriculum learning. The EED‐MATD3 method employs various exploration strategies, and the controller and power distributor are treated as two agents. Through centralized training and decentralized execution, a robust cooperative control strategy is realized. The performance of the proposed algorithm is verified in an LFC model of Zhuhai Tandang Island, an island microgrid in the China Southern Power Grid.

As the proton exchange membrane fuel cell (PEMFC) is a nonlinear, time-varying, multiple-input multiple-output system, an advanced controller with strong robustness and adaptability is required for controlling PEMFC stack temperature and achieve a high operation efficiency. In this paper, a data driven optimal controller is proposed for controlling the stack temperature, which is based on large-scale deep reinforcement learning. In addition, a new deep reinforcement learning algorithm termed curriculum guidance strategy large-scale dual-delay deep deterministic policy gradient (CGS-L4DPG) algorithm is proposed for this controller. The design of this algorithm introduces the concepts of the curriculum guidance strategy and imitation learning, and its inclusion improves the performance and robustness of the proposed controller. The simulation results show that, taking advantage of the high adaptability and robustness of CGS-L4DPG algorithm, the proposed controller can more effectively control the PEMFC stack temperature than existing control algorithms. © 2021 The Authors. IET Renewable Power Generation published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology