Study On Assembly Line Balancing Optimal Models And Algorithms

In order to achieve economic and smooth production, Assembly line is used to manufacture mass products; Assembly line balancing is a process to reach equilibrium among productivity, utilizing rate of facility and market demands.This thesis deals with methodology of assembly line balancing on the ground of mass production and with small lot sizes product variety. This method is based on the optimal mathematical model and modern AI. From single mode 1 (including determined assembly line and stochastic assembly line to mixed model, from straight shaped line to U-shaped line, from single objective to multiple objectives, from resource free to resource constraints assembly line, the model and methodology is developed to balance the assembly lines systemically.Investigating models on various objectives and various types of assembly line balancing problems. Optimal model on I-type and II-type goal assembly line balancing problem is proposed in this paper; meanwhile, optimal model on U-line is proposed. In addition, models on multi-objective assembly line balancing problem and resources-constraints assembly line balancing problem are set up. Comparing to the traditional linear programming model, the flexibility of the linear programming model proposed in this paper is good; it overcomes the limitations of traditional mathematical model. The numbers of variables and constraints is reduced dramatically. The optimal models provide some good methodology and tools for production manager.Systematically designs AI algorithms for on various objectives and various types of large-scale assembly line balancing problems. In terms of the goal of assembly line, genetic algorithms for minimized work station numbers, minimized cycle time, minimized smoothness index and maximized similarity index are established; in terms of various assembly line, genetic algorithms for determined, stochastic, mixed-model, multi-objectives and resources constrained assembly line are proposed. For fitness function of the chromosomes, a fitness function to realize continuous improvement for the algorithms is selected; simultaneity, heuristic jointly with random ways to generate initialized populations; a kind of simple, practical and unique principle for gene encoding and decoding is developed, especially on U-line, the gene is encoded with signal showing the assigning direction, which leads to efficient assembly line balancing. Additionally,...