The disassembly is a fundamental basis in converting End-of-Life (EOL) products into useful components. Related research becomes popular recently due to the increasing awareness of environmental protection and energy conservation. Yet, there are many opening questions needed to be investigated, especially the efficient coordination of different-level decisions under uncertainty is a big challenge. In this paper, a novel integrated stochastic disassembly line balancing and planning problem is studied to minimise the system cost, where component yield ratios and demands are assumed to be uncertain. In this work, machine specificities are considered for task processing, such as price, ability, and capacity. For the problem, a two-stage non-linear stochastic programming model is first constructed. Then, it is further transformed into a linear formulation. Based on problem property analysis, a valid inequality is proposed to reduce the search space of optimal solutions. Finally, a sample average approximation (SAA) and an L-shaped algorithm are adopted to solve the problem. Numerical experiments on randomly generated instances demonstrate that the valid inequality can save around 11% of average computation time, and the L-shaped algorithm can save around 64% of average computation time compared with the SAA algorithm without a big sacrifice of the solution quality.

拆卸是将报废（EOL）产品转换为有用组件的基本基础。由于对环境保护和节能意识的提高，相关的研究最近变得很流行。然而，有许多悬而未决的问题需要研究，特别是在不确定性下有效协调不同级别的决策是一个巨大的挑战。本文研究了一种新颖的集成随机拆卸生产线平衡与规划问题，以最大程度地降低系统成本，在这种情况下，组件的成品率和需求都不确定。在这项工作中，要考虑机器的特殊性以进行任务处理，例如价格，能力和容量。针对该问题，首先构建了两阶段非线性随机规划模型。然后，它进一步转化为线性公式。在问题性质分析的基础上，提出了一个有效的不等式，以减少最优解的搜索空间。最后，采用样本平均逼近（SAA）和L形算法来解决该问题。在随机产生的实例上进行的数值实验表明，与SAA算法相比，有效不等式可以节省大约11％的平均计算时间，L形算法可以节省大约64％的平均计算时间，而不会牺牲解质量。