Using evolutionary algorithm based on hybrid probability distribution to solve TSP with area partition

As the traditional evolutionary algorithms for large-scale TSP(Traveling Salesman Problem) produce so huge amount of paths vectors with random that the slow and premature convergence is nearly inevitable, this paper presents a novel evolutional algorithm based on hybrid probability distribution(EABHPD) with area partition strategy. The fundamental idea is to use evolutionary algorithm twice. Firstly the large-scale TSP is divided into several small-scale TSP, then each sub-TSP can be solved with EABHPD. With EABHPD, the rules of mutation are the combination of Gaussian probability distribution, Cauchy probability distribution and t probability distribution. This designed algorithm can get a good compromise of the desired precision and computation cost, it also can avoid the premature convergence problem of the common evolutionary algorithms. Besides, the efficiency of our approach is manifested by the preliminary simulation experiment.