Waste collection has always been a major research area in waste management. It plays an important role in social development and environmental sustainability. However, the past research often makes great efforts to formulate dedicated models to some specific waste collection applications, and relatively speaking, fewer efforts have been devoted to relevant method development. Inspired by these issues, in this work, we first develop a more general two-echelon multi-objective location routing problem model (2E-MOLRP) in consideration of the inherent similarities of many realistic waste collection applications. In the model, various commonly-seen and potential costs are classified in a straightforward way and different objectives can hence be flexibly defined to satisfy different requirements. Furthermore, to solve the model, an improved non-dominated sorting genetic algorithm with directed local search (INSGA-dLS) is proposed. In order to validate its effectiveness, experiments are conducted in comparison with existing representative metaheuristics and the results show that our proposed algorithm can achieve better performance even without using local search. Also, we prove that the specially-designed directed locate search is able to further improve our algorithm’s performance significantly in experiments.

废物收集一直是废物管理的主要研究领域。它在社会发展和环境可持续性方面发挥着重要作用。然而，过去的研究经常为建立用于某些特定废物收集应用程序的专用模型而付出很大的努力，相对而言，用于相关方法开发的工作较少。受这些问题的启发，在这项工作中，我们首先考虑到许多实际废物收集应用程序的内在相似性，首先开发了一个更通用的两级多目标位置路由问题模型（2E-MOLRP）。在该模型中，以简单明了的方式对各种常见成本和潜在成本进行了分类，因此可以灵活定义不同的目标以满足不同的需求。此外，要解决模型，提出了一种改进的有向局部搜索的非支配排序遗传算法（INSGA-dLS）。为了验证其有效性，与现有的代表性元启发式算法进行了比较，结果表明，即使不使用局部搜索，我们提出的算法也可以实现更好的性能。此外，我们证明了经过特殊设计的定向定位搜索能够在实验中进一步显着提高算法的性能。