Application Of Adaptive Ant Colony-Genetic Algorithm Based On Similarity Degree And Population Entropy

Under the market steep competition, the production scale of the manufacturing industry increases day by day. In the actual production, unresolved problem often is the large-scale scheduling problem based on thousands of machines, thousands of orders each month. Because of the essence of the production scheduling question that is combined the optimized question, but the existing production scheduling algorithms all concentrates in the small scale scheduling domain, if directly used in these large scale production scheduling, the overwhelming majority algorithms, in the storage space and the computing time, is all unable to accept, Based on the present researches, in view of the large-scale this characteristic, how to propose some effective algorithms, are close to the practical application of the scheduling algorithms, has become a new hot spot about the production scheduling domain research in recent years.The tasks of the paper is presented as follows:A self-adaptive ant colony-genetic algorithm based on similarity degree and population entropy is proposed in the paper, which is used to solve the large-scale scheduling problems. The initial population is divided into four kinds of small population, the exploration, the development, the exploration development, and the retained. According to various populations' function, the different genetic parameter and the different evolution strategy is used in different population in this algorithm. In the algorithm, this article establishes the population entropy with similar degree, which can reflect population's multiplicity from the most direct viewing. And the population's scale is adjusted dynamically by the population factor correlated to the population entropy, guaranteeing the population's multiplicity, strengthening the algorithm's parallelism. Simultaneously in order to make full use of the feedback information in the algorithm, considered the character of ant colony algorithm and retained population synthetically, the ant colony algorithm is used to seek superiorly in the retained population, which not only may prevent the most superior loss, but may speed up the entire algorithm the convergence rate.In the above algorithm the exploration population is an algorithm's core, it shoulders the responsibility for seeking and developing problem's solution space, the algorithm in it not only needs to have the quickly speed of seeking solution, but the algorithm itself robustness must be strong. In the algorithm similarity degree is a very essential and very important parameter, therefore this paper abandons the original self-adapted genetic algorithm's consistent strategy that algorithm do the mutation after the cross. The similarity degree is took as the watershed which decides to carry on either cross or mutation, the threshold value of similarity degree is adjusted dynamically through a function correlated the evolution number correlation, this kind of strategic change enhances exploration population's exploration ability, also enhances the algorithm whole to seek the superior ability.Experimental study: This article, in view of the Job-Shop scheduling question, the Flow-Shop scheduling question and some automobile spare part limited liability company's AAM production line, has designed and realized a scheduling algorithm experiment system. And then the inprovement algorithm is applied to these questions; the results show improvement algorithm was feasible and effective.