Abstract: Aiming at the randomness characteristics of disassembly task time due to a lot of uncertain factors in the actual disassembly line,a stochastic model of disassembly line balancing problem was constructed which was based on the disassembly model for part precedence relation diagram definition and takes the number of workstations,the balance index and the stability index as the optimization objective under the premise of meeting the cycle time constraint,and an local neighborhood genetic algorithm based on Pareto was proposed.A decoding method for random operation time was designed.The global search of the population was realized by two kinds of crossover operations,and a local search strategy combining depth neighborhood and breadth neighborhood was constructed to expand the scope of local search and improve local search ability.The proposed algorithm was applied to solve two large-scale disassembly cases,and the result indicated the effectiveness of the proposed algorithm and the validity of the improvement strategy.By taking a television disassembly case including 27 tasks as an example,the practical application of the proposed model and algorithm was identified by analyzing the application process and result.

Key words: disassembly line balancing problem, stochastic task times, genetic algorithms, local neighborhood

摘要： 针对实际拆卸线存在的不确定性因素导致作业时间具有随机性的特点,采用零件优先关系图定义的拆卸模型,以工作站数目、平衡性指标、稳定性指标为优化目标,考虑工作站等效作业时间满足节拍时间约束,构建了随机型拆卸线平衡问题模型,并提出一种基于Pareto占优的局部邻域遗传算法。在该算法中,设计了一种面向随机作业时间的解码方法,通过两种交叉操作实现种群的全局搜索,并构造了深度邻域和广度邻域相结合的局部搜索策略,以扩大局部搜索的范围并提高局部寻优能力。通过对两个大规模算例的测试与对比,验证了所提算法的优越性和改进策略的有效性。最后,将模型和算法运用至27项任务的电视机为拆卸实例,通过分析该随机型拆卸线平衡优化的具体应用过程与结果,说明了所建模型与算法的实用性。

关键词: 拆卸线平衡问题, 随机作业时间, 遗传算法, 局部邻域