An Improved Real-value Adaptive Genetic Algorithm

The improvement problem of genetic algorithm is studied based on real coding.To the shortcomings that: the search is inefficient and it is easy to premature convergence,the parameter adjusting problem of genetic algorithm is discussed.The adaptive crossover probability and adaptive mutation probability are proposed,considering the influence of every generation to algorithm and the effect of different individual fitness in every generation.Three testing functions are used to validate the algorithm.The results thaw that the ultimate value,the average algebraic sum and the convergence probability all obtain the preferable values.