The Research And Reality Of A Optimizational Mixed Strategy Genetic Algorithm

As an optimal method, genetic algorithm has obvious advantages, which is based on the nature selection and genetic transmission mechanisms such as high collateral, stochastic, self-reliance. It has good advantages in finding the best global answer of optimization problem. Genetic algorithm works with coding form, uses only object function information and uses probabilistic transfer regular, genetic algorithm has operational parallelism and cares for every individual quantity in evolutionary populations namely the quanlity of problem solution. Therefore, genetic alogrithm is used on several fields and becomes one of powerful tools to solve the global optimization questions. Although genetic algorithm has several character, as a optimization method, it exists still its limitation, especially existed the phenomenon of premature, slow rate of convergence and the deficiency of local search ability, which affect the performance of genetic algorithm. In order to overcome the limitation of genetic algorithm and to improve the efficiency and convergence rate of genetic algorithm solving function optimization questions, a new improved genetic algorithm-optimizational mixed strategies genetic algorithm (abbreviated to OMSGA) is proposed by improving crossover operator. Two type of crossover operator - single parent single pair single offspring crossover operator (SPSPC) and single parent two pairs of single offspring crossover operator (SPTPC) are presented in the paper. The character of two crossover operators is that only single parent generates one offspring in procedure of generating offspring population, and generates different shape new individual by performing randomly crossover operator to single parent. Single parent crossover ensures that a new generation has the basic charater of feasible solution, and improves diversity of individual and search ability to solution fields. Single crossover operator can make any one parent to generate another new individual by finite order genetic crossover.The algorithm implements automatically to adjust mixed probability of crossover stratesy for every generation population by the strengthen and weaken strategy of mixed probability. The intilization of evolution, the algorithm sets the same mixed probability of pure crossover strategy for every generation population, which happens to continuous changes with evolution of population. The strengthen and weaken strategy of mixed probability makes population in evolution procedure to select the crossover strategy with more higher probability, and it makes the possibility boosted that the algorithm selects crossover strategy with more higher mixed probability for population in every generation, and crossover strategy with more higher mixed probability is the crossover strategy can generate more better individuals in previous evolution procedure, so that it improves the rate of converence of the algorithm.(?) The more important significanceis to propose a new improved strategy and make genetic algorithm to deal effeciently with optimization questions, and make the application of genetic algorithm more deeper. By programming by MATLAB language and going the test that is carried out to six traditional functions, the results of test shows the algorithm is excellent to the single crossover strategy of genetic algorithm and multi-point orthogonal crossover operation genetic algorithm (MPOCGA) in rate of converence for dealing with function optimization questions.