The Study Of Genetic Algorithm Optimization Based On TSP

In the field of artificial intelligence, research in the search process automatically access and accumulated knowledge of the search space to control and adaptive search process to be the optimal solution for general search has been remarkable subject. TSP is combinatorial optimization problems in the field of a typical problem, Its statements is very simple, but the solution is very difficult and has been proved to be NP-complete problems. Genetic algorithms are from biological mechanism of natural selection and evolution of the overall situation to develop probabilistic search algorithm.Among the many ways to resolve TSP, the genetic algorithm has many advantages which other methods have not. for small and medium-sized TSP. it can search the best result, for large-scale TSP, it can be similar to the optimal solution.This paper compared the genetic algorithm common select operator, cross operator, mutation operator of genetic algorithm and determine the effect of tourn-elite selection, partially-mapped crossover, displacement mutation and insertion mutation operator based optimization. Since tourn-elite selection is too concentrated in a few good individual phenomenon, Make a new method of selection: different parents tourn-elite selection. Since inversion mutation and inserted mutation have the respective advantages and disadvantages of variation, Make a new mutation: inversion-insert mixed mutation.Compared the orginal algorithm and the algorithm It porved by random location cities to test the result is good or bad after the same generations. The result shows that the improved method can produce better results. It also reached another conclusion: the same number of cities TSP problems, and use the same method if using ring arranged cities can be the optimal result, with random arranged cities may not be the optimal result, so ring arranged TSP test does not have commonality. All the tests used by FLASH, all code used by the Action Script2.0 language.