The Improved Genetic Algorithm And Applicatioo In TSP Problem

In this paper,the because the evolutionary process of the basic genetic algorithm is not obvious for each generation and the gaps between fitness values, this paper has a deep reach and puts forward two improved methods,and the same time to solve the TSP problem (traveling salesman problem). It plays a guiding role of theory for the effective utilization of network routing and resource.Firstly,for the problem that the gap between the fitness values is not obvious, the paper puts forward a new method to calculate the degree of adaptation. The main consideration of the algorithm is a steep change curve index, and to get the length of the path from each population,and to get the transformation of the exponential function, the function value further discretization to obtain the different fitness values between each population,and to show the optimal unit of the population process, and then to transmit to the next generation for the next generation population optimization.Secondly, for each evolution of the basic genetic algorithm is not obvious, this paper changes the operation sequence on the basic genetic algorithm,and to improve the proportional selection operator,the population of the present generation through mutation and crossover is combined with the population of the former generation, so it can easily lead to the optimal population of the new generation;the improvement of the executive scale operator method, it consists of the present generation and the former generation to get a new population.To sort the new population by length, thereby to reduce the worse individuals possibility enter into the next generation by the roulette of the basic genetic algorithm.To improve for a detailed calculation process on the single point crossover operator, The main purpose is to increase the number of intersection, thus increasing the possibility of the new individuals;to improve the method of mutation, it mainly is the mutation before and after the populations which is combined to produce new population. Finally,by improving the method of the above two kinds of improved genetic algorithm, the convergence speed and the optimal solution has been greatly improved, so as to achieve the expected effect, and considering the parallelism of the genetic algorithm, considering local optimization effect is good of ant colony algorithm and the parallel information transmission of parallel algorithm, to use the double error method,and the paper has carried on the exploratory experiment.