Collaborative vehicle-drone distribution network optimization for perishable products in the epidemic situation

Abstract

The collaborative vehicle-drone distribution network (CVDDN) optimization problem incorporates vehicle-drone collaboration mechanism, site selection, product perishability, and epidemic impact into the optimization framework. And it is formulated as a bi-objective mathematical model to minimize the total cost and the value loss for products distribution. An efficient two-phase hybrid heuristic algorithm based on the improved K-means clustering and the extended Non-dominated Sorting Genetic Algorithm-II (ENSGA-II) is proposed to solve the investigated CVDDN optimization problem. The improved K-means clustering algorithm is employed to find an effective location strategy for “contactless” distribution sites (DS) and act as a base for collaborative vehicle routing problem with drones (VRPD). The ENSGA-II combines the NSGA-II algorithm framework and flexible tabu search rules to ensure a large effective search and fast parallel calculations, which generates a real-coded solution space to find the hybrid vehicle-drone delivery routes for VRPD. Various experimental instances tests show that the ENSGA-II outperforms other algorithms. An empirical case study of Chengdu city in China indicates the efficiency of our proposed optimization model and solving approach. Sensitivity analysis is conducted to identify the impact of various parameters on the CVDDN optimization problem.