Lecture title：Application of reinforcement learning in maintenance and spare parts inventory optimization

Time: 15:00-16:30 p.m. on Friday, December 2, 2022

#Tencent conference: 622-389-115

Reported by: Professor Zhou Yifan

Brief introduction of the reporter:

Zhou Yifan, professor and doctoral advisor, served as the executive director of the Industrial Engineering Branch of the China "Double Law" Research Association, the director of the Reliability Branch of the China Operation Research Association, and the director of the China Creation Association. Served as the Arrangement Chair of IEEE PHM International Conference and the Session Chair of several international conferences. He presided over two general programs of the National Natural Science Foundation of China, one youth fund of the National Natural Science Foundation of China, one doctoral program fund of the Ministry of Education, and two open funds of key laboratories of Jiangsu Province. Published more than 40 SCI/EI retrieval papers, including 16 SCI papers published by the first author/corresponding author in Reliability Engineering and System Safety, Applied Mathematical Modeling, IEEE Transactions on Reliability, Computers&Industrial Engineering and other journals. He was selected as the "Top Ten Tutors" in the "My Favorite Tutors" of Southeast University.

Introduction to the report:

Multi-agent reinforcement learning and deep reinforcement learning are important methods for solving large-scale Markov decision processes that have emerged in recent years. It provides a new idea for the maintenance optimization of large multi-component system and the inventory optimization of multi-warehouse system based on Markov decision process. This report introduces the principles and methods of reinforcement learning, multi-agent reinforcement learning and deep reinforcement learning. It shows its application in maintenance optimization and inventory optimization. The advantages and disadvantages of reinforcement learning and heuristic algorithm in solving such problems are compared and discussed, as well as the different application scope of the two algorithms. Finally, the future research directions are discussed.

(Undertaken by: Department of Management Science and Logistics, Digital Economy Innovation Research Center of Yangtze River Delta Research Institute, Scientific Research and Academic Exchange Center)