The Application Of The Evolutionary Computation For Optimal Problem

In many fields of science and technology, industries and practice etc, there are a lot of problems can be converted into the kind of mathematical model of function optimization. Evolutionary algorithm is one of the most effective algorithms for hard optimization and multi-objective optimization problems, which are attached more and more importance to. This paper studies evolutionary algorithms for single objective and multi-objective optimization.In the course of evolution, there may be premature convergence. It is mainly because the super individual has appeared in the population, according to a certain selection strategy, this individual will occupy the absolute predominance in the population soon, and this makes the algorithm to converge exploitation too early. In this paper, the solution about this question is to adjust the fitness function of the super individual to control it selection probability, or to increase the mutation probability of the individual to increase the diversity of the population. Meanwhile, different population scale and selection strategy play a very important role in the algorithm function.In this paper different crossover and mutation operator are used, and combined with simulated annealing. The ability to make the genetic value of individual stocks is more effective in maintaining diversity of population, under the traditional crossover algorithm is easy to fall into local optimal solution of the defect; and can guide the genetic algorithm in optimal neighborhood search, thereby enhancing the speed of optimization.The TSP is studied and a novel genetic approach for TSP without fixed Starting point and end point is proposed. To make the propose algorithm effectively, the solution representation is simplified first. As a result, a simplified encoding scheme is presented, and an efficient decoding scheme is designed. Then based on the specific encoding scheme, the efficient crossover and mutation operators are proposed. It can always generate valid tour. In order to enhance its ability of exploration, a local search scheme is integrated into the crossover operator. Based on these, a novel and effective genetic algorithm for TSP is presented and its convergence to global optimal solution is proved. The simulation results show the effectiveness of the proposed algorithm.