A Multiobjective Hybrid Genetic Algorithm for TFT-LCD Module Assembly Scheduling

作     者：Chou, Che-Wei Chien, Chen-Fu Gen, Mitsuo 

作者机构：Natl Tsing Hua Univ Dept Ind Engn & Engn Management Hsinchu 30013 Taiwan Fuzzy Log Syst Inst Fukuoka 8200067 Japan 

出 版 物：《IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING》 (IEEE Trans. Autom. Sci. Eng.)

年 卷 期：2014年第11卷第3期

页      面：692-705页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 0811[工学-控制科学与工程] 

基　　金：National Science Council, Taiwan [NSC 96-2221-E-007-048, NSC 102-2622-E-007-013, NSC 103-2622-E-007-002, NSC 102-2221-E-007-057-MY3] Hsinchu Science Park [102A26] Toward World Class University Project from the Ministry of Education [102N2075E1] Japan Society of Promotion of Science (JSPS) Grants-in-Aid for Scientific Research Funding Source: KAKEN 

主　　题：Multiobjective genetic algorithm scheduling thin-film transistor-liquid crystal display (TFT-LCD) TOPSIS variable neighborhood descent 

摘      要：The thin-film transistor-liquid crystal display (TFT-LCD) module assembly production is a flexible job-shop scheduling problem that is critical to satisfy the customer demands on time. On the module assembly shop floor, each workstation has identical and non-identical parallel machines that access the jobs at various processing velocities depending on the product families. To satisfy the various jobs, the machines need to be set up as the numerous tools to conduct consecutive products. This study aims to propose a novel approach to address the TFT-LCD module assembly scheduling problem by simultaneously considering the following multiple and often conflicting objectives such as the makespan, the weighted number of tardy jobs, and the total machine setup time, subject to the constraints of product families, non-identical parallel machines, and sequence-dependent setup times. In particular, we developed a multiobjective hybrid genetic algorithm (MO-HGA) that hybridizes with the variable neighborhood descent (VND) algorithm as a local search and TOPSIS evaluation technique to derive the best compromised solution. To estimate the validity of the proposed MO-HGA, experiments based on empirical data were conducted to compare the results with conventional approaches. The results have shown the validity of this approach. This study concludes with a discussion of future research directions. Note to Practitioners-Because of short product lifecycles, cycle time reduction and on-time delivery are crucial in high-tech industries such as the TFT-LCD and semiconductor manufacturing. To address these needs in real settings, a novel multiobjective hybrid genetic algorithm (MO-HGA) was developed, hybridizing with a variable neighborhood descent (VND) algorithm as a local search and TOPSIS technique to select the best compromised solution for the TFT-LCD module assembly scheduling problem. Experiments have shown practical viability of this approach. Future studies can be done to exten