Article ID: | iaor19921024 |
Country: | Japan |
Volume: | 26 |
Issue: | 12 |
Start Page Number: | 1427 |
End Page Number: | 1434 |
Publication Date: | Dec 1990 |
Journal: | Transactions of the Society of Instrument and Control Engineers |
Authors: | Abe Keisuke, Tsukiyama Makoto |
Keywords: | scheduling, artificial intelligence, programming: mathematical, vehicle routing & scheduling |
As consumer needs become more and more complex there is an increasing need for flexible transportation systems at the distribution level, in the same way as flexible manufacturing systems are needed at the production level. In this paper, the authors propose an efficient method for road transportation scheduling, which combines a knowledge base system with optimization algorithms. They consider the problem of scheduling road transportation among a set of cities. This is a large scale combinatorial problem and constraints are not fixed but flexibly changed. The basic approaches to this problem are considered as follows. 1)Mathematical programming approach, 2)knowledge engineering approach. It is not guaranteed that a mathematical programming method can be applied when constraints are changed. As for the knowledge engineering approach it may be difficult to acquire useful knowledge. To overcome these problems authors have developed a scheduling method which combines knowledge based systems with a conventional optimization algorithm. The knowledge based subsystem describes the problem constraints and selects feasible solution candidates. Such a space of feasible solutions is searched by the optimization algorithm. The proposed method makes it possible to get near optimal solutions for large scale problems in a short time. The schedule can be easily modified by a planner in the process of generating solution candidates. When problem constraints are changed, only the knowledge base is affected, without any impact on the software. [In Japanese.]