| On 2007-10-10 at 14:00:00 (Brussels Time) |
Abstract
Although production scheduling is one of the more applied fields in optimization, most research is done in far simplified problems. This helps to learn about heuristics, metaheuristics and exact methods, about problem structures and many other things. But it does not help to solve the problems industry faces. We can see a trend to consider constraints related to these real-world problems. However, almost no research is to the combination of many of these constraints. A hybrid flexible flow line with eligible unrelated machines, precedence constraints, sequence depending setup times, release dates and time lags results in a complex problem, closely related to the problems in many industrial sectors. We try to find good solutions for problems of a reasonable size, within reasonable time. Therefore metaheuristics are a logical choice. We have developed Genetic Algorithms with distinct solution representations. The performance is compared to heuristics and for the smallest instances to the performance of a Mixed Integer Programming model. Currently we are working on an Iterated Gready algorithm, concentrating on Local Search, to see how this compares to the Genetic Algorithms. Some initial results will be given.
Keywords
scheduling, local search
