Applying Simulated Annealing with Matching Improvement to Solve Fuzzy Multi-Objective Unrelated Parallel Machine Scheduling Problems

• Main Authors: Ruei-Chi Li, 李瑞琪
• Other Authors: Chiuh-Cheng Chyu
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/15367624230401430799

碩士 === 元智大學 === 工業工程與管理學系 === 98 === The research investigates several strategies that are incorporated into simulated annealing (SA) to solve unrelated parallel machine scheduling problems with two maximization fuzzy objectives – total completion time (makespan) satisfaction and average tardiness satisfaction. These strategies include matching-based decoding schemes, acceptance probability rules, and random and fixed weighted vectors for objectives. In particular, a two-phase matching decoding scheme is devised to cope with the search direction on the two objectives and improve the solution quality. Two acceptance probabilities of neighborhood solutions based on the following rules are considered: fitness values of solutions, and the number of dominated solutions in the archive. An experiment was conducted to evaluate the performance of SA with different strategies, using three instance sets of moderate to large sizes generated by a method in the literature. The experimental results indicate that (1) the two-phase decoding scheme that uses max-min matching first and Hungarian method second will significantly improve proximity quality; (2) random weighted direction search is better than fixed weighted, (3) dominance-based SA will produce enhanced diversified local efficient solutions.