Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
[Display omitted]
•Dynamic maintenance strategy optimized for single machine based on MDP.•State transition probability and maintenance operations change with status.•Integrated model established to solve conflict between maintenance and production.•Maintenance strategy updated from joint decision model for best benefits.
Semiconductor production has characteristics of reentry, uncertainty, and high level of equipment integration, which are different with general machining processes. It involves many different types of equipment which are highly coupled by re-entrant wafers and hundreds processing steps. Therefore, the reliability of equipment and throughput of the process should be balanced by reasonable scheduling on production and maintenance. It is unreasonable only focusing on the maintenance of single device and ignoring the interaction between machines as previous literatures in this field. This paper addresses the integrated decision problem of production and maintenance for a deteriorating multi-machine system (semiconductor production line). In this paper, a two-step strategy is presented to solve the joint decision problem while considering the overlaps between maintenance and production time. First, we develop a dynamic maintenance plan which includes maintenance time points and actions choices for each single machine based on a Markov decision process (MDP). Next, an integrated decision model of production scheduling and maintenance plan is put forward to balance an entire semiconductor production line when maintenance times conflict with production processing. Choices of delaying maintenance or production are decided through maximizing total economic benefits of production line. The dynamic maintenance strategy is researched and compared with other arbitrary methods. A typical semiconductor production line (Mini-Fab model) is tested with the presented approach. Simulated results from MATLAB and EM-Plant software show that the maintenance method presented can improve system benefits and usability of key equipment. This paper contributes to fill the research gaps mentioned above and provides a more practical production and maintenance control model for industrial practitioners.