A Selective Population-based Algorithm with Multi-Perturbative Operators for Traveling Salesman Problems

A selective population-based algorithm with multi-perturbative operators is proposed in this article for the classical combinatorial optimization problem: traveling salesman problem. The proposed algorithm is composed of two important components: perturbative operator set and operator management method. Two types of operators are contained in the perturbative operator set: (1) five basic operators that are most commonly used by metaheuristics in solving traveling salesman problem, (2) five composite operators that are proposed based on the execution sequence of five basic operators. A scoring table method is proposed in this article to manage the operators in the set. Thirty-seven datasets from TSPLIB are adopted to demonstrate the performance of the proposed algorithm. From the experimental results, two phenomena are observed: (1) composite operators obtain better solutions than basic operators on all datasets while consume more running time than basic operators especially on larger datasets, (2) scoring table mechanism enables the algorithm to obtain better solutions on 34 of 37 datasets. Additionally, among eight state-of-the-art metaheuristics, the proposed algorithm outperforms six of them on all datasets used while outperforms the remaining two on 9 of 11 datasets and 21 of 23 datasets respectively.