State Key Laboratory of Advanced Electromagnetic Engineering and Technology, and School of Electrical and Electronic Engineering, Huazhong University of Science and Technology, Wuhan, China
Auto-generated profile, based on publicly available information
Yujun Lin
State Key Laboratory of Advanced Electromagnetic Engineering and Technology, and School of Electrical and Electronic Engineering, Huazhong University of Science and Technology, Wuhan, China
This paper addresses the home energy management (HEM) problem for a large number of residential houses, which can be regarded as a high-dimensional optimization problem. To cope with the high-dimensional issue, the concept of the aggregator is utilized to reduce the state and action space. And a two-stage deep reinforcement learning (DRL) based approach is proposed for the aggregators to track the schedule from the superior grid and guarantee the operation constraints. In the first stage, a DRL control agent is set to learn the optimal scheduling strategy interacting with the environment based on the soft-actor-critic (SAC) framework and generate the aggregate control actions. In the second stage, the aggregate control actions are disaggregated to individual appliances considering the users’ behaviors. The uncertainty of the EV charging demand is quantitatively described by the driver’s experience. An aggregate anxiety concept is introduced to characterize both the driver’s anxiety on the EV’s range and uncertain events. Finally, simulation studies verify the effectiveness of the proposed approach under dynamic user behaviors, and the comparisons also show the superiority of the proposed approach over the method mentioned in benchmarks.
16 June 2025 in IEEE Transactions on Sustainable Energy
Numerous small-scale energy storage systems (ESSs) are distributed throughout the power system and have the potential to be aggregated for power regulation. In this context, distributed power tracking control with aggregated capacity estimation is proposed for distributed ESSs. The proposed distributed control is free of the measure for total power while achieving accurate total power tracking. In addition, power allocation is performed among ESSs considering their types, SoC levels, and power/energy capacities. To estimate the dispatched power ranges of ESSs, an event-triggered distributed power capacity estimator is proposed with a relatively low communication burden. The power reference for the power tracking control is limited and therefore the reliable operation of the control is ensured. The proposed control and estimator are implemented in a discrete-time fashion and only rely on the sparse communication requirement. A detailed theoretical analysis is conducted to verify the equilibrium and convergence of the proposed approaches. Simulations based on the IEEE 33-bus system demonstrate the effectiveness of the proposed distributed power tracking control and the capacity estimator.
In this study, the home energy management problem, which can be regarded as a high-dimensional optimization problem, for numerous residential houses, is addressed. The concept of the aggregator is utilized to reduce the state and action space and to handle the high dimensionality. A two-stage deep reinforcement learning (DRL)-based approach is proposed for the aggregators to track the schedule from a superior grid and guarantee the operation constraints. In the first stage, a DRL control agent is set to learn the optimal scheduling strategy interacting with the environment based on the soft-actor-critic framework and generate the aggregate control actions. In the second stage, the aggregate control actions are disaggregated to individual appliances considering the users' behaviors. The uncertainty of an electric vehicle's charging demand is quantitatively expressed based on the driver's experience. An aggregate anxiety concept is introduced to characterize the driver's anxiety on the electric vehicle's range and uncertain events. Finally, simulations are conducted to verify the effectiveness of the proposed approach under dynamic user behaviors, and comparisons show the superiority of the proposed approach over other benchmark methods.