Abstract: To better reflect the uncertainty of actual disassembly task time,a multi-objective U-shaped mathematical model of disassembly line balancing problem was formulated by considering the stochastic operation times.A Pareto based hybrid artificial fish swarm algorithm was proposed for overcoming the deficiency that failure to balance each objective of traditional method in solving multi-objective problems.For reducing the repeated search in parallel foraging of fish swarm,a serial foraging way adopting the self-adaptive visual field was introduced,and the introduction of Pareto non-inferior solutions set realized the diversity of fish swarm optimization results.A simulated annealing operation was operated on the sequence results which could avoid the local optimum.In addition,the crowding distance was employed as an evaluation mechanism to filter and retain the elite solutions.Furthermore,the elite solutions were added to the next iteration of population which speeded up the convergence rate.The proposed algorithm was applied to one certain printer disassembly instance including 55 disassembly tasks.The effectiveness and superiority were validated through the comparison of proposed algorithm with basic artificial fish swarm algorithm and simulated annealing algorithm.

Key words: U-shaped disassembly line balancing, stochastic operation times, multi-objective optimization, artificial fish swarm algorithm, simulated annealing algorithm, Pareto set

摘要： 为更好地反映实际拆卸作业时间的不确定性,建立了考虑随机作业时间的多目标U型拆卸线平衡问题的数学模型,并针对传统方法求解多目标问题时求解结果单一、无法均衡各目标等不足,提出一种基于Pareto解集的多目标混合人工鱼群算法。算法采用自适应视野串行觅食方式,以减少并行觅食时出现重复搜索现象,并根据多目标拆卸序列之间的支配关系得到Pareto非劣解集,实现了鱼群寻优结果的多样性。对鱼群觅食得到的拆卸序列进行模拟退火操作,增强了算法跳出局部最优的能力。采用拥挤距离机制筛选非劣解,实现了拆卸序列的精英保留,进而将非劣解添加到下次迭代的种群中,加快了算法的收敛速度。将所提算法应用于具有55项任务的某打印机拆卸实例,经与基本人工鱼群算法、模拟退火算法对比,验证了所提算法的有效性和优越性。

关键词: U型拆卸线平衡, 随机作业时间, 多目标优化, 人工鱼群算法, 模拟退火算法, Pareto解集