A Genetic Algorithm for the Dynamic Capacitated Arc Routing Problem

抄録

Route optimization problem has been an indispensable role in the transportation and logistics industry. In this paper, the dynamic capacitated arc routing problem (DCARP) system is determined as the sequence of static CARP within the planned time horizon. Compared to the static problem that all demands are known provisionally, DCARP has deterministic demands which remain unknown until the appearing time comes. The new demand appears must be updated and rearranged while carrying out the current servicing tasks. Since both CARP and DCARP are NP-hard problems, we develop a genetic algorithm (GA) to DCARP which is periodically solved as CARP after new demand is updated. Due to no available instances to test for the DCARP, we randomly generate DCARP instances from CARP benchmark instances. The extensive numerical experiments show that the proposed GA can obtain comparable results. In the future, we can apply our GA on real world cases.