Using micro genetic algorithm for solving scheduling problems
Job Shop Scheduling Problem (JSSP) and Timetable scheduling are known to be computationally NP–hard problems. There have been many attempts by many researchers to develop reliable scheduling software, however, many of these software have only been tested or applied on an experimental basis or on a s...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Thesis |
| 语言: | English |
| 出版: |
2005
|
| 主题: | |
| 在线阅读: | http://eprints.utm.my/id/eprint/35024/1/TayChengSan%20MFKE2005.pdf |
| 标签: |
添加标签
没有标签, 成为第一个标记此记录!
|
| 总结: | Job Shop Scheduling Problem (JSSP) and Timetable scheduling are known to be computationally NP–hard problems. There have been many attempts by many researchers to develop reliable scheduling software, however, many of these software have only been tested or applied on an experimental basis or on a small population with minimal constraints. However in actual model JSSP, the constraints involved are more complicated compared to classical JSSP and feasible schedule must be suggested within a short period of time. In this thesis, an enhanced micro GA, namely micro GA with local search is proposed to solve an actual model JSSP. The scheduler is able to generate an output of a set of feasible production plan not only at a faster rate but which can generate a plan which can reduce the makespan as compare to those using manual. Also, in this thesis, the micro GA is applied to the timetabling problem of Faculty of Electrical Engineering Universiti Teknologi Malaysia which has more than 3,000 students. Apart from having more students, the faculty also offers various different type s of specialized courses. Various constraints such as elective subjects, classrooms capacity, multiple sections students, lecturer, etc have to be taken into consideration when designing the solution for this problem. In this thesis , an enhanced micro GA is proposed for timetable scheduling in the Faculty to overcome the problems. The enhanced micro GA algorithm is referred to as distributed micro GA which has local search to speed up the scheduling process. Comparisons are made with simple GA methods such that a more optimal solution can be achieved. The proposed algorithm is successfully implemented at the Faculty meeting a variety of constraints not achievable using manual methods. |
|---|