Abstract
Abstract
This paper focuses on the cold chain logistics and proposes a multi-objective Vehicle Routing Problem (VRP) model that seeks to minimize the total cost of cold chain logistics and maximize the fairness of employees’ workloads. This model first incorporates carbon emissions into the cost structure, and also discretely calculates vehicle departure times by introducing time-dependent factors. Specifically, an algorithm named NSGA-ALNS is developed to solve the proposed model. Different from the well-known genetic algorithmic frameworks that include the selection, crossover, and mutation steps, the proposed algorithm utilizes the population evolution strategy of NSGA-II for global search, employs Adaptive Large Neighborhood Search (ALNS) as a local search operator for route refinement, and further introduces a population diversity strategy to reduce duplicated individuals. Empirically, comparing and analyzing the Pareto frontiers of NSGA-ALNS, NSGA-II, and MOALNS on the Solomon dataset and Gehring & Homberger’s instances, we clearly demonstrate that NSGAALNS improves the convergence and the extent of the solution set.