TSP Problem Solution Based on Improved Genetic Algorithm

PMX operator can be utilized while TSP is resolved by Genetic Algorithm. Its defect are (1)match region is strictly restricted for PMX crossover operator. PMX cannot realize crossover while partial genes are the same and others are different. (2) PMX crossover operator is a double point crossover, which is bad for the heredity of excellent genes. Aiming to these shortcoming about PMX, EPMX (Extend PMX) is proposed by this paper, firstly this operator can select a crossing position arbitrarily, match region is before the crossing position, crossover region is after the crossing position; secondly using match region to find position mapping rules; thirdly genes of crossover region are exchanged according to position mapping rules; at last crossover region is exchanged. Discrete Bet Wheel of Select operator and Dmutation mutation operator are proposed in this paper, they are improvement effectively of select operator and mutation operator respectively. TSP problem is resolved efficiently by this improvement GA. Experiment tell that with the same environment max-optimization solution appears in 75 generation of new algorithm rather than 48, and put off prematurity phenomena, that max-fittness value is increased from 87 to 100, and convergence, veracities are improved contrasted with traditional method.