The Research On Genetic Algorithm Based On Adaptive And Chaos

Genetic Algorithm (GA) has been studied extensively in recent years. Compared with other intelligent algorithms, Genetic Algorithm possesses the advantages of suppleness, universality, robustness, parallelism and extensive use. And it gradually becomes one of the key technologies of intelligent algorithm in the 21st century.Although Genetic Algorithm had been widely applied in many fields and get a lot of research findings,It still exist a lot of deficiencies needed to be improved and perfected, which is slow convergence, premature convergence probably, can not converge to the global optimal solution with the probability. In the actual application, strengthening local search ability, speeding up the convergence, fast reaching global optimal solution, these problems must be synthetically considered. Around these problems, the main contributions are presented as follows.1. In order to rebalance convergence rate and global convergence, this thesis proposes an improved and adaptive Genetic Algorithm(IGA). In this algorithm, using a switch to spread initial stocks in global space of solution; utilizing adaptive crossover and mutation operators to increase the various adegrees of stocks and reach global optimization. Since standard genetic algorithm can not converge to the global optimal solution, improved Genetic Algorithm applies a stratagem, where reserves the optimal solution before selection operator. And certified this solution is probability convergence basing on Markov Chain. At last, gives the application of this improved algorithm in PID Controller and indicates having the stronger searching ability.2. Based on the hybrid algorithm, this thesis puts forward a hybrid genetic algorithm combined with the chaos (CGA). Since chaos has internal and special randomness, utilizes chaos'random sequence to increase the various degrees of stocks. This algorithm improves the local searching capacity and quickly reaches the global optimal solution. It is applied in two functions optimization: Banana and Schaffer functions, and compared with the above improved and adaptive genetic algorithm. Numerical results show that this hybrid algorithm can effectively alleviate the problem of premature convergence and greatly increase the speed of global convergence. At the same time, gives the comparison with the other proposed chaos genetic algorithm, the result shows CGA can search better accurate the global optimal solution.