Design a Robust Logistics Network with an Artificial Physarum Swarm Algorithm

Abstract

The robust optimization of logistics networks can improve the ability to provide sustainable service and business sustainability after uncertain disruptions. The existing works on the robust design of logistics networks insisted that it is very difficult to build a robust network topology, and this kind of optimization problem is an NP-hard problem that cannot be easily solved. In nature, Physarum often needs to build a robust and efficient topological network to complete the foraging process. Recently, some researchers used Physarum to build a robust transportation network in professional biological laboratories and received a good performance. Inspired by the foraging behavior of natural Physarum, we proposed a novel artificial Physarum swarm system to optimize the logistics network robustness just on a personal computer. In our study, first, the robustness optimization problem of a logistics network is described as a topology optimization model based on graph theory, and four robustness indicators are proposed to build a multi-objective robustness function of logistics network topology, including the relative robustness, the betweenness robustness, the edge robustness and the closeness robustness. Second, an artificial Physarum swarm system is developed to simulate the foraging behavior of a natural Physarum swarm to solve this kind of complex robust optimization problem. The proposed artificial Physarum swarm system can search for optimal solutions by expansion and contraction operations and the exchange of information with each other through a self-learning experience and neighbor-learning experiences. The plasmodium of Physarum forms the edges, and the external food sources simulate the logistics nodes. Third, an experimental example is designed on the basis of Mexico City to verify the proposed method, and the results reveal that the artificial Physarum swarm system can help us effectively improve the logistics network robustness under disruptions and receive a better performance than natural Physarum. The article may be helpful for both theory and practice to explore the robust optimization in logistics operation and provide engineers with an opportunity to resist logistics disruptions and risk loss by a novel artificial intelligence tool.