时间：9月18日（本周四）上午10:30 - 11:30

地点🪪🚡：舜德楼北510会议室

报告人：日本爱知工业大学，大野胜久教授

报告题目：Optimal Control of JIT Based Production and Distribution System by Neuro-Dynamic Programming and Comparison with Pull Systems

报告摘要：

Just-In-Time (JIT) production systems are successfully implemented in many factories in the world. This lecture deals with a multi-stage JIT based production and distribution system with stochastic production and demand. The JIT production systems use a production ordering kanban and a withdrawal kanban and require an input buffer for parts and an output buffer for products in each stage. An optimal control problem without using kanbans of the JIT based production and distribution system is analyzed for minimizing the expected total cost per unit time. Note that almost papers on JIT production systems and/or pull systems deal with production systems using one kind of kanban and having one input or output buffer in each stage. The problem can be formulated as an undiscounted Markov decision process (UMDP), but it is very difficult to exactly solve it due to the curse of dimensionality. In recent years, several algorithms in the field of reinforcement learning, or neuro-dynamic programming (NDP), have been developed to overcome the curse of dimensionality. This lecture presents a new algorithm of NDP, which is called the simulation-based modified policy iteration method (SBMPIM), for solving the optimal control problem. The SBMPIM and existing algorithms of NDP are numerically compared with an optimal control policy of a single-stage production line computed by the MPIM of UMDP.

The kanban system is the first pull system and then several pull systems have been proposed. This lecture presents a simulation optimization algorithm that finds optimal parameters of pull systems such as Kanban, Base Stock, CONWIP, Hybrid and Generalized Kanban. These optimally set pull systems are numerically compared with suboptimal control policies of three-stage production and distribution systems computed by the SBMPIM.

报告人简介：

Katsuhisa Ohno is a professor at Aichi Institute of Technology in Toyota, Japan. His research interests include supply chain management, Just-In-Time production and distribution systems, queueing networks, Markov decision processes and neuro-dynamic programming. He received the B.S., M.S., and Ph.D. degrees in operations research from Kyoto University （京都大学） in 1964, 1966 and 1973, respectively. He was a member of the Editorial Advisory Board of Transportation Research and Associate Editors of J. Operations Research Soc. Japan. He is now Editor In Chief of International Journal of Logistics and SCM Systems, a member of the Editorial Board of European Journal of Operational Research and a member of the Editorial Board of International Journal of Production Research.

报告人主要简历：

• April 1966 - March 1967:   Industrial Engineer, Toyota Motor Company

• April 1967 - December 1967:   研究员, Osaka University （大坂大学）

• January 1968 - March 1984:    副教授, Kyoto University （京都大学）

• April 1984 - March 1986:    教授, Konan University （甲南大学）

• April 1986 - March 1994:    教授, Nagoya Institute of Technology （名古屋工业大学）

• April 1994 - Present:    教授, Aichi Institute of Technology （爱知工业大学）

欢迎感兴趣的师生准时参加👩🏻‍🎨。