An Anti-fragile Optimization Approach for Integrated Energy System Based on Safe Reinforcement Learning

Integrated energy system (IES) is one of the energy fields that have attracted much attention globally, and the high penetration of distributed energy sources makes it difficult to regulate the IES, which has a high load density and high security requirements, and the improvement of its anti-fragile capability is a key problem to be solved urgently. In this paper, an anti-fragile optimal scheduling method for IES based on safe reinforcement learning is proposed. Firstly, a mathematical model of IES safety constraints including physical constraint information such as economic objectives and unit constraints is constructed. Second, the model is transformed into a constrained Markov decision process, and a special scheduling safety cost is designed to circumvent unsafe actions of agents. Subsequently, the scheduling safety cost is embedded into the decision-making process, and a safe reinforcement learning model is constructed based on the constrained Markov decision-making process to carry out example testing. The results show that the proposed method can solve the IES anti-fragile optimization problem better.