Research On Stochastic Two-sided Mixed-model Assembly Line Balancing Problem

Two-sided assembly line is mainly used to assemble large and complex products, due to the complexity of the products, more and more plants begin to adopt two-sided assembly line. Compared with the traditional assembly line, two-sided assembly line can shorten the length of the assembly line, reduce the product of off-line time, save the cost of tooling, fixtures and materials handling, etc. In order to adapt to the diversity of market, the mode of production has been changed a lot, vary from single type to varieties. Although the higher automation it is, the active of people can’t be taken place by automation machine, manual assembly in the most of industry is normal. For the manual mixed-model assembly line, the working time may be random due to some factors. The Stochastic Two-sided mixed-model assembly line Balancing problem is closer to the actual condition, it’s worth studying. Two-sided assembly line balancing problem is complicated than single side assembly line balancing problem, when the tasks were assigned to the work-station, need to consider the constraint of precedence relation between the tasks assigned to the same mated-station, because the constraint may cause a "waiting-time". Complicated constraint will increase the complexity of modeling and solving of the problem. Hence, it’s need to find a new method to solve the Stochastic Two-sided mixed-model assembly line Balancing problem, and it is value for the industry’s production.To reduce the complexity of solving the stochastic two-sided mixed model assembly line balancing problem, a two-stage procedure of combining a genetic algorithm with simulation analysis is proposed. Firstly, according to the characteristics of Two-sided mixed-model assembly line Balancing problem, a simplified mathematical programming model is formulated ignoring the constraint of precedence relation between the tasks assigned to the same mated-station, and a set of solutions is obtained by using a sequenced-based, combined coding genetic algorithm. Based the set of solutions, a simulation model that take all the constraints of the problem into account is developed in the eM-Plant software. The optimal solution (or suboptimal solution) of the problem is obtained through simulation analysis and evaluation.Finally, taking a refrigerator assembly line’s actual problem as the case study, and using this two-stage method to solve the actual problem. The result shows that it can meet the needing of assembly line, and improve assembly line balancing effect. In conclusion, the analysis of an illustrative example shows that the complexity of solving the problem can be greatly reduced and a satisfied solution can be obtained with the proposed method.