A deep reinforcement learning approach based energy management strategy for home energy system considering the time-of-use price and real-time control of energy storage system
Shengtao Xiong, Dehong Liu, Yuan Chen, Yi Zhang, Xiaoyan Cai
Abstract
To enhance the flexibility of the home load optimization dispatching strategy and ensure the safe operation of the energy storage system, an optimization dispatching strategy for home energy management system (HEMS) based on time-of-use pricing and real-time energy storage system control is proposed. Firstly, a HEMS dispatching model is established with the constraints of dispatchable load and energy storage system working status, aiming to minimize the total cost for home users. This dispatching strategy controls the energy storage system charging and discharging behavior based on time-of-use pricing and real-time battery state of charge, which helps to reduce the power cost for home users and ensure the safe operation of the battery. Then, the optimal dispatching problem of HEMS is modeled as a Markov decision process (MDP) and solved by a deep reinforcement learning algorithm called soft actor-critic (SAC). The example results verify the effectiveness and superiority of the proposed method compared with other benchmark methods, which the system cost can be reduced by 15.87% at least.