Application Of Immune Genetic Algorithm With Learning Ability On Job-Shop Scheduling

With economic of market development, many sequences and small quantity become the focal point of the market which manufactory racing to control. In this way, manufactories should be asked to range sequences rationally, take advantage of resource, shorten time limit for a project and reduce cost of producing. So people paid attention to the JSPmore and more.JSP belonging to NP-hard problem, is the hardest solving problem in classic optimization problem. GA is applied in the optimization of JSP widely, because it has characteristics that it can be used generally and the algorithm is simple.But the characteristic of strongly general using leads to badly flexibility. For the algorithm is simple though it can guarantee its entire convergence, it will essentially degenerate in part. Now many experts dedicate to solving GA which is applied in JSP, and it is a bottle-neck problem.In this paper, machine-learning theory is applied in IGA, and initial population which have higher average fitness is built by memorization function of machine-learning. Advanced vaccine is guaranteed by learning function which can pick up vaccine from the best individual. This made the process of evolvement developing along the better direction.The theory of steady-state reproduction and the more advanced mode of coding are induced into the algorithms in this paper. The capability of the algorithms is improved greatly.Knowledge base is built by SQL Server 2000. The process and the result of optimum are stored in different ways. The efficiency of the searches is improved by all of these operations.Fitness function is optimized by Exp function. This made individual of higher fitness possessing more selected probabilities and lower fitness possessing less selected probabilities.Presently, the algorithms are applied in "The optimized plat of Job-shopmanagement" successfully. Furthermore the algorithms are running in real job-shop scheduling successfully.