Parallel Binary Cat Swarm Optimization with Communication Strategies for Traveling Salesman Problem

Abstract

This paper studies the Parallel Binary Cat Swarm Optimization (PBCSO) algorithm and its application to the Traveling Salesman Problem (TSP). It investigates the performance of PBCSO with two evolution strategies employing neighborhood merge comparison and dynamic adaptation, aiming to optimize population fitness. This parallel mechanism randomly divides initial solutions into few groups, and shares the information in various groups following every fixed iteration. In this way, the defects of the original BCSO premature convergence and easy to fall under the local optimal search space can be significantly reduced. We have conducted repeated tests on 23 functions of three types. The results demonstrate that the proposed PBCSO can solve the optimization problem more specifically. It can maximize the diversity of the population, so that the optimal solution is most likely to be obtained, which is easy to break through the limitations of local convergence to achieve the global optimal. TSP is a classic NP-hard problem, which has been extensively studied in the literature. This article shows that PBCSO can be successfully used in the analysis of TSP problems and has broad application prospects. PBCSO was simulated on Matlab, and the result proved the feasibility and effectiveness.