Hybrid K-means and Particle Swarm Optimization for symmetric Traveling Salesman Problem

The Traveling Salesman Problem (TSP) is well-known established scheduling problems. We propose a novel method for the TSP using the divide-and-conquer strategy. We employ K-means algorithm to find the city clustering and then solve a sequence of sub-city in a given order by Particle Swarm Optimization (PSO). The PSO is modified by incorporating genetic algorithm operators, namely mutation, so that it can the escape from the local optimum. The performance of proposed method is tested against a number of instances from the TSPLIB. Results demonstrate the effectiveness of the proposed method. Moreover, the novel method gives better results in the standard TSP problem than the exist algorithm.