Research And Implementation Of Distributed Scheduling System Based On Mixed Local-search Algorithm

As manufacturing companies expand production scale, production processes become increasingly complex, companies need the aid of computer enterprises to arrange product procedure, improve production efficiency. But in the real production scheduling process, faced with complex processing environment and resource constraints,we need to design a effective algorithm to solve the scheduling problem.Job Shop scheduling problems are combinatorial optimization problems,there are many algorithm to solve this kind of problem local-search algorithm is a kind of algorithm which is most widely used, of which the genetic algorithm is the most successful method. In this paper, through the research of multi-objective flexible job shop scheduling problem,improve the traditional genetic algorithm for its shortcomings, and add the simulated annealing algorithm and immune mechanisms, finally, design a job-shop scheduling algorithm based on mixed local-search algorithm.The algorithm selected part of the chromosome in the initial stage, dispersed chromosomal distribution through judgments of chromosome similarity, improve the algorithm's global search capabilities; and the rest of the chromosome using the immune mechanism, make use of some characteristics of known information to generate high-quality chromosomes, laying a good foundation for the subsequent evolution. At the same time,for the uniform speed of evolution of the algorithm,designing a adaptive cross and mutation probability, the fitness value of different chromosomes given different probabilities of crossover and mutation, and the probability will decreases with the evolution generation increase.speed up the convergence of algorithm in the final stage. In the selection operation,algorithm introduce simulated annealing mechanism and the immune mechanism combined strategy to avoid the algorithm fall into local minimum which making algorithm can not find better solutions, and save the best chromosome of every generation. Finally, we prove the validity of algorithm.Then construct a distributed job shop scheduling system around the shop scheduling algorithm,so that it can be applied in practice. The system is based on Web service+EJB technical specifications and service-oriented architecture. The paper introduces the whole system architecture and implementation principles, and the server side, Web module, the client's design was described in detail. Finally, the paper test performance of the algorithm and the distributed system through testing and evaluating the algorithm many times, and compared with the standard genetic algorithm and other mixed algorithms.the experiment show that ,the improved mixed-local search algorithm have better capabilities in global search . Finally, testing the distributed system in the clients which is based on different technical standards, and achieved good results.