A multi-product job shop scenario utilising Model Predictive Control

Author(s)

Sprodowski, Tobias ^{[1, 2]} ; Sagawa, Juliana Keiko ^[3] ; Maluf, Arthur Sarro ^[3] ; Freitag, Michael ^{[1, 2]} ; Pannek, Jurgen ^{[1, 2]}

Affiliation

^[1] Univ Bremen, Fac Prod Engn, Badgasteiner Str 1, D-28359 Bremen - Germany

^[2] Univ Bremen, BIBA Bremer Inst Prod & Logist GmbH, Hsch Ring 20, D-28359 Bremen - Germany

^[3] Univ Fed Sao Carlos, Dept Prod Engn, Rod Washington Luis Km 235, BR-13565905 Sao Carlos, SP - Brazil

• Document type: Journal article
• Source: EXPERT SYSTEMS WITH APPLICATIONS; v. 162, DEC 30 2020.
• Web of Science Citations: 0

Abstract

Multi-product manufacturing scenarios today have to face many challenges considering external factors such as availability of resources or attending product demands and internal factors such as adjustment of buffer levels or utilisation of workstations. In this paper, a multi-product job shop consisting of workstations coupled by an unidirectional material flow with different production routings is considered. The existing model based on the bond graph technique is adapted to an optimal control problem, allowing the applicability of the Model Predictive Control scheme. Concerning performance criteria, two different objective functions are defined: the first aims for predefined processing frequencies of the workstations and the second one takes into account product demands. Both approaches were examined in simulations showing that a steady state is achieved in terms of stable buffer levels and processing frequencies. (C) 2020 Elsevier Ltd. All rights reserved. (AU)

• FAPESP's process: 19/12023-1 - Dynamic models for the control of job shop production systems
• Grantee: Juliana Keiko Sagawa
• Support Opportunities: Regular Research Grants