Parallel simulation-based optimization on scheduling of a semiconductor manufacturing system

As an important and challenging problem, the scheduling of semiconductor manufacturing is a hot topic in both engineering and academic fields. Its purpose is to satisfy production constraints on both production process and resources, as well as optimizing some performance indexes like cycle-time, movement, etc. However, due to its complexities, it is hard to describe the scheduling process with a mathematical model, or to use conventional methods to optimize its scheduling problem. A Simulation approach is proposed to optimize the scheduling of a semiconductor manufacturing system, i.e. a simulation-based optimization (SBO) approach. Because the high computational cost of SBO approach could hinder its application in the real production line, a parallel/distributed architecture is discussed to improve its efficiency. Using genetic algorithm (GA) as an optimization algorithm, the proposed parallel-SBO based scheduling approach for semiconductor manufacturing system is tested for its feasibility and effectiveness.