On Dec. 14, 2018, invited by Professor Leyuan Shi, director of the Department of Industrial Engineering and Management, Professor Baris Tan from Koc University, Turkey paid a visit to the College of Engineering. The two parties introduced their schools with a hope of seeking future cooperation opportunities. Afterwards, Prof. Tan gave a talk entitled “Data-Driven Control of a Production System by Using Marking-Dependent Threshold Policy”.
Professor Baris Tan giving the talk
Baris Tan is a Professor of Operations Management and Industrial Engineering and the Vice President for Academic Affairs and Provost at Koç University, Istanbul, Turkey. His areas of expertise are in design and control of production systems, supply chain management, and stochastic modelling. He is the author of many publications including two edited books and serves as the manufacturing area editor of Flexible Services and Manufacturing Journal and Design and Manufacturing area editor of IISE Transactions.
In his talk, Prof. Tan introduced that his study is motivated by the need of developing effective data-driven control methods for production systems. His aim is answering the following research questions: How can we model inter-event time accurately by using the data; How can we control production by using partial information (markings) obtained from a production system; and How can the control parameters be determined by using the real-time data?
In order to analyze the performance of a production system controlled with partial information, they introduce the Marked Markov Arrival Process (MMAP) framework to model a system that generates different signals, referred as markings. They then propose a marking-dependent threshold policy to control the system. An analytical method that is based on a matrix analytical approach is developed to analyze a production/inventory system with MMAP information and demand arrivals and a partially observable production time process modeled as a MMAP. A mathematical programming formulation is used to determine the optimal thresholds of the control policy based on the matrix geometric model of the system. They then present a joint simulation and optimization (JSO) approach to determine the parameters of the threshold policy by using the shop-floor data collected from the system.
They show that using a marking-dependent control policy together with a JSO approach that determines the policy parameters works effectively as a data-driven control method for manufacturing.
After the lecture, the audience raised many questions and Prof. Tan answered them accordingly.