A Simulation with PSO Approach for Semiconductor Back-End Assembly

Abstract

This paper studies a dynamic parallel machine scheduling problem in a hybrid flow shop for semiconductor back-end assembly. The facility is a multi-line, multi-stage with multi-type parallel machine group, and orders scheduled with different start time. As a typical make-to-order and contract manufacturing business model, to obtain minimal manufacturing lead time as main objective and find an optimal assignment of production line and machine type by stage for each order as main decisions. Nevertheless, some production behavior and conditions increase the complexity, and including order split as jobs for parallel processing and merged completion for shorten lead time. Complying quality and traceability requirement so each order only can be produced from one of qualified line(s) and machine type(s) and all jobs with the same order can only be produced in same assigned line and machine type with stochastic processing time. Lead time is counted from order start time to completion, including sequence dependent setup times. As a NP-hard problem, we proposed a simulation optimization approach, including an algorithm, particle swarm optimization (PSO) to search optimal assignment which achieving expected objective, a simulation model to evaluate performance, and combined with optimal computing budget allocation (OCBA) to reduce replications. It provides a novel applications using simulation optimization for semiconductor back-end assembly as a complex production system.