Hybrid estimation of distribution algorithm with multiple subpopulations for semiconductor manufacturing scheduling problem with limited waiting-time constraint

This paper considers a semiconductor manufacturing scheduling problem (SMSP), subjected to all the practical constraints such as limited waiting time, machine status, different process time on different machines, setup time and arrival time in wafer fabrication facilities (fabs) of semiconductor manufacturing industry. A hybrid estimation of distribution algorithm with multiple subpopulations (HEDA-MS) is proposed to solve SMSP effectively within several specified minutes for an online scheduling requirement. An empirical study simulates eight scenarios from practical data to compare the performance of HEDA-MS and GA, not only to minimize the makespan, but to make total exceeded of limited waiting time into zero. For all the scenarios, the proposed HEDA-MS obtains a smaller makespan than GA with less total exceeded limited waiting time.