The self-organization genetic algorithm based on the mutation with cycle probabilities

First, a cycle mutation genetic algorithm (CMGA) is designed by simulating the evolutionary rule of the earth creature found by paleontologists in the paper. Then, according to some phenomena of the population genetics, an improved cycle mutation genetic algorithm (ICMGA) is schemed by mended the selection operator of CMGA. Last, 22 functions are tested by ICMGA and other evolution algorithms in the experiments. The results show that exploration and exploitation of ICMGA are better than those of other algorithms and ICMGA is not sensitive to the initial population distribution. By the statistical analysis of the evolutionary course, it has been found that ICMGA has the self-organization ability which is similar to that discussed in the theory of complex system.