A Dual-Archive Memetic Algorithm with Motion Prediction for Solving Dynamic Multiple Traveling Salesmen Problem

The Dynamic Multiple Traveling Salesmen Problem (DMTSP) is a generalization of the Traveling Salesman Problem. DMTSP deals with multiple salesmen and moving targets, making it more consistent with real-life scenarios. Solving DMTSP is very challenging due to the min-max objective and dynamic characteristics. In this paper, we propose a Dual-Archive Memetic Algorithm with Motion Prediction (DAMA-MP) to solve this problem. The DAMA-MP maintains two archives to save elite solutions and profiteer solutions. The parent generation is selected from the archives for crossover to inherit the strengths of elites and profiteers. The local intensification is performed on the elite archive to enhance algorithm development capabilities. Moreover, a motion prediction (MP) module is introduced during the evaluation stage to estimate the fitness of the solutions. Finally, the experimental results show that DAMA-MP has outstanding performance. Especially at large scales, DAMA-MP has faster convergence speed and better convergence value than the other algorithms.