In this paper, we investigate the Pollution Routing Problem in dynamic environments (DPRP). It consists in determining the routing plan of a fleet of vehicles supplying a set of customers, while minimizing the traveled distance and \(CO_2\) emissions. The dynamic character of the problem is manifested by the occurrence of new customer demands when the working plan is in progress. Consequently, the planned routes have to be adapted in real time to include the locations of the new customers. In order to efficiently manage the trade-off between the two considered objectives, a new vector evaluated evolutionary algorithm augmented with an exploitation phase and hyper-mutation is proposed. This combination aims to reinforce the refinement of compromised solutions, and to speed up adaptation after the occurrence of a change in the problem inputs. An experimental study is conducted to test the proposed approaches on mono-objective and bi-objective test problems, and against well known approaches from the literature. The obtained results show that our proposal performs well and is highly competitive compared with the competing meta-heuristics.

在本文中，我们研究了动态环境（DPRP）中的污染路由问题。它包括确定为一组客户提供服务的车队的路线计划，同时最小化行驶距离和\(CO_2\)排放。问题的动态特征表现为工作计划进行时新客户需求的出现。因此，计划的路线必须实时调整以包括新客户的位置。为了有效地管理两个考虑的目标之间的权衡，提出了一种新的向量评估进化算法，该算法增加了利用阶段和超突变。这种组合旨在加强对妥协解决方案的改进，并在问题输入发生变化后加快适应。进行了一项实验研究，以测试所提出的单目标和双目标测试问题的方法，并与文献中众所周知的方法进行对比。