Genetic algorithm (GA) is a random search method based on the biological evolution. It roots in the natural selection mechanism of the survival of the fittest. It is simple, robust and can be used in the parallel computation. Its global search performance is very excellent, but GA also has some shortcomings such as slow convergence speed, easy to be premature, bad local search capability. And it can't meet the requirements of search speed and solution precision simultaneously. In order to solve the complex, multi-variable optimal problems, great deals of research works have been done in this thesis.The influence of each stage of natural evolution on species is analyzed. The conception of the probability of constituting schemata (PCS) is presented considering the impact factors on the survival of schema are discussed. We set up a relationship between parameters selection and GA performance theoretically by analyzing the influence of the PCS on the performance of GA. Experiments are conducted to test the performance of GA from several aspects.In order to maintain the diversity of population (DP) in binary-coded genetic algorithm, a new adaptive mutation operator is proposed after the effect of gene frequency (GF) to the DP. The mutation probability in some certain locus is a function of the GF. Simulation results show the validity of the proposed method.To overcome the demerit of crossover operator while operating without direction, a new crossover operator is proposed .Combining the idea of simplex method, this new operator enhances the ability of GA to finding the global optimal.A climbing operator is proposed to enhance the climbing ability of GA. Experiments are made to verify the efficiency of the climbing operator in advancing the diversity of GAAt last, the improved genetic algorithm proposed in this thesis is applied to the base station placement planning of CDMA network. |