Automobile Mixed-model Assembly Balancing Via Stochastically Ordered Cellular Automata Algorithm

Automobile industry undergoes a development from standardization to diversification. Withthe intensified market competition, more varieties of models are produced to meet differentlevels and personalized user demands. Automobile assembly is the final procedure of the wholeproduction process, and the degree of workload balance has a great impact on the product qualityand production efficiency. Therefore, in order to meet various product demands from market andimprove the production performance, it’s necessary to carry out an elaborate study on theautomobile assembly line balancing problem and design a fast and effective algorithm.In this thesis, the mixed-model automobile assembly line balancing problem is solved bythe stochastically ordered cellular automata algorithm via four steps. First, A mathematicalmodel is formulated with the objectives of minimizing the production cycle time and increasingthe balance efficiency. Subsequently, the given problem is represented by cellular automata,where tasks are considered as mobile particles, workstations as fixed grid nodes, the mechanismof balancing assembly lines transported as evolution rules. Third, the initialization iscomplemented by heuristic selection and task assignment to speed up the initial feasible solutionand improve its performance. Forth, after designing the generalized swapping process of tasksbetween any two stations, an orderly execution rule is proposed according to non-decreasingnumber of swapped tasks, so as to converge fast to the current best solution, and stochasticselection rule among stations and tasks are constructed, with the purpose of avoiding beingtrapped into local optimum.Illustrated by a car engine assembly line and benchmark problems, the cellular initializationand stochastically ordered cellular evolution rules are validated to be correct and applicative.Relevant experiment results show that the proposed stochastically ordered cellular automataalgorithm gets near optimal solutions or optimal solutions within a short time for different scaleproblems.