An Advanced Genetic Algorithm for Traveling Salesman Problem

By analyzing the deficiency of traditional genetic algorithm in solving the Traveling Salesman Problem, an improved genetic algorithm is proposed for TSP. In this paper, the ordinal real-number encoder is used for chromosome encoding and ordered crossover operators is advanced that utilizes local and global information to construct offspring. In order to guarantee global convergence, heuristic knowledge and self-learning is employed for mutation. Then, a city network which contains 31 city nodes is employed to test the algorithm. The simulation result of MATLAB shows that the proposed method can get feasible result with a higher convergent rate and success rate than existing heuristics.