Logo image
Optimizing Preventive Maintenance Using Genetic Algorithm and Discrete-Event Simulation
Thesis

Optimizing Preventive Maintenance Using Genetic Algorithm and Discrete-Event Simulation

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

Abstract

Optimization Preventive maintenance scheduling PM window Genetic Algorithm Discrete-Event Simulation
The study established the way to solve a real complex system regarding preventive maintenance scheduling problem in semiconductor manufacturing system, particularly in finding the appropriate start time of PM within PM window such that production loss due to maintenance activity can be minimized. To deal with multidimensional search space, meta-heuristics such as genetic algorithm and particle swarm optimization were introduced. Discrete event simulation was embedded into meta-heuristics algorithm for solving optimization problem. According MANOVA (Multivariate ANalysis Of VAriance) of five-performance indicators, -throughput rate, work in process (WIP), cycle time, equipment utilization and manpower requirement-, at the 0.05 level, it pointed out that a genetic algorithm-discrete event simulation (GA-DES) and a particle swarm optimization-discrete event simulation (PSO-DES) approach performs better than resource leveling and reference. For practical problem in this study, under objective of minimizing manpower, GA-DES perform better than PSO-DES; however under objective of maximizing throughput, there was not enough statistical evidence at the 0.05 level to conclude that GA-DES and PSO-DES were different each other. Contributions of this study include: 1. Identifying the best arrangement of the start time of PM within PM window. 2. Providing a way to optimize PM schedules for a complex system by utilizing meta-heuristics and discrete event simulation, simultaneously.

Metrics

1 Record Views

Details

Logo image