Based On Estimation Of Distribution Algorithm Solving Hybrid Flow-shop Scheduling Problem

In recent years, with the gradual development of society, the importance of science and technology in the production of social development become increasingly prominent. By using the progress and development of science and technology to improve productivity, reduce production costs and improve the competitiveness of enterprises and more attention in various fields. The so-called shop scheduling is to allocate available resources to meet the enterprise’s normal, orderly and rapid output, which is to allocate the workpiece, machining equipment and its relationship with the order of time, in order to achieve maximum use of the equipment and improve production efficiency. Therefore, the research shop scheduling problem has important scientific research and production value.With the rise of intelligent algorithms for solving some of the complex shop scheduling problems provides an effective tool to further expand the space for shop scheduling problem research. Estimation of Distribution Algorithms are emerging in recent years, an evolutionary algorithm, it will establish the probability model and the sample is introduced into the process of evolution, evolutionary algorithms to replace traditional analog operation of crossover and mutation process of natural evolution. The algorithm to guide the search process by probability model, this can prevent blindness and randomness of chromosomal rearrangements, and thus effectively improve the efficiency of the search, to quickly and accurately solve many complex optimization problems by traditional evolutionary algorithms are difficult to resolve.In this paper, the characteristics of flow shop scheduling problem, the application of distributed estimation algorithm for flow shop scheduling problem is solved. Because there is a certain lack of traditional estimation of distribution algorithms, such as a single fixed probability vector, prone to premature convergence phenomenon, in order to improve the convergence speed, this article will introduce ideas simulated annealing algorithm to estimation of distribution algorithm which will accept the simulated annealing after mutation probability is applied to select the dominant populations, a new estimation of distribution algorithm based on simulated annealing to improve, while the algorithm is optimized in order to reduce the performance and efficiency of the dependence of the initial population, this paper first uses a heuristic NEH algorithm to generate a better solution, and then generate the initial population using random sampling methods. Through the distribution of improved estimation algorithm, and thus the characteristics of hybrid flow shop scheduling problem, this paper presents a method of coding the algorithm to solve the problem of population initialization method and probabilistic models, algorithms and select update method probabilistic model. Finally, use the classic scheduling problem and a factory workshop scheduling and a scheduling system, through the examples of simulation tests validate the algorithm’s effectiveness and superiority. The results were analyzed data obtained with this algorithm to optimize the quality is good, stable performance, small affected by the initial value, and strong global search capability superiority.