Watch Video
11 30, 2026
Speaker: Lu Jin,The University of Electro-Communications
Inviter: 胡庆培 研究员
Title: Symmetry-Aware Deep Reinforcement Learning for Joint Load Allocation and Maintenance: From Two-Unit to Multi-Unit Systems
Language: Chinese
Time & Venue: 2026.11.30 09:30-10:30 南楼N402
Abstract: In load-sharing systems, the load assigned to each unit affects its deterioration, so load allocation and maintenance decisions are best considered jointly. This talk focuses on a system of two identical units with continuous load allocation. We formulate the problem as a Markov decision process and show how interchanging the units relates their optimal decisions. We incorporate this policy symmetry into proximal policy optimisation (PPO) through an auxiliary loss function. Numerical experiments show that symmetry-aware PPO achieves higher simulated returns and closer agreement with the benchmark policy obtained by value iteration for a discretised model. The talk concludes with a brief look at ongoing work on extending this framework to multi-unit systems.
Copyright@2008, All Rights Reserved, Academy of Mathematics and Systems Science, CAS
Tel: 86-10-82541777 Fax: 86-10-82541972 E-mail: contact@amss.ac.cn
京ICP备05002806-1号 京公网安备110402500020号