Logo image
Machine-Process Grouping Considering Setup and Waiting Times Using Particle Swarm Optimization
Thesis

Machine-Process Grouping Considering Setup and Waiting Times Using Particle Swarm Optimization

Dedy Suryadi
Masters, 國立清華大學, 工業工程與工程管理學系
2008

Abstract

machine-process grouping setup time waiting time PSO
This research focuses on developing a machine-process grouping algorithm in which taking waiting time and setup time into the optimization’s objective. The method used to find the best grouping is Particle Swarm Optimization (PSO). Thus, the algorithm is named SWPSO (Setup & Waiting, PSO). A particle representation is created with dimension length equals to (M + P), i.e. sum of number of machines and recipes to be processed. The position boundaries are between 0 and min{M,P}. For the fitness function in PSO, estimations are used for both setup and waiting time. In particular for waiting time, there is a proportion of non-processing time in the makespan which needs to be found. Through repeated simulation, it is found out that the actual makespan is 1.73 times the ideal one. Thus, there is a proportion of non-processing time as much as 0.73. The grouping result is compared with the result from previous research (Cheng, 2008). It is called SGRAM (Setup, Dendogram) because it considers only setup time and uses dendogram to find the best grouping. Comparison is also made with the original grouping, most number of groups, and no grouping. For most performance measures under different lot release policies, SWPSO shows better performance. In the comparison of two dispatching policies for SWPSO grouping, CR shows better performance than FIFO in all but average throughput. Furthermore, in additional cases where the numbers of machine and recipes to be processed are perturbed, SWPSO consistently shows better performance than or at least equal to SGRAM for most performance measures.

Metrics

1 Record Views

Details

Logo image