Research Of Some Scheduling Problems Base On Hybrid Intelligent Algorithm

It has a very important practical significance to seek excellent scheduling algorithm for theimprovement of today’s social productive forces. This research, which is supported by NationalNature Science Foundation of China and Guangdong Nature Science Foundation, is based on thispoint to extend.Firstly, the paper presents the background of shop scheduling problem, the description andclassification of shop scheduling problem, the feature of shop scheduling problem, researchmethods of shop scheduling problem and the current research status. Then this paper introducesthe basic theory of the genetic algorithms and the electromagnetism-like mechanism method. Dueto the lack of the enlightening, the stability of genetic algorithm in solving this kind of problems isrestricted. The information sharing mechanism and enlightening of electromagnetism-likemechanism method are strengthened comparing with genetic algorithms. Hence, this paperproposes a new hybrid algorithm, so called EMGA, which combines electromagnetism-likemechanism method and genetic algorithm. Further more, a kind of random key, which is base onEMGA, is designed. The EM of the algorithm is improved and the validity of the improvement isvalidated by comparing experiment.This paper uses the framework of EMGA to solve job shop scheduling problem and flexiblejob-shop scheduling problem. For the job shop scheduling problem, we use the framework ofEMGA to design a hybrid algorithm. A kind of code and genetic operators, which are based onrandom key, are designed. The effectiveness of the combination of EM and GA and the hybridalgorithm are showed by experiments.For the flexible job-shop scheduling problem, considering of wider search space and morecomplex, this paper not only designs a kind of code and genetic operators, which are based onrandom key and the problem, but also proposes an improved Kacem strategy, which is aim atstrengthening the quality of initial population. And a kind of improved chaotic local searchstrategy and machine adjustment strategy are also designed to enhance the search capability of thealgorithm. At last, the validity of the improvements are showed by experiments. The validity ofthe hybrid algorithm is also demonstrated by experiments.Finally, the work is summarized and the direction of future research is discussed.