Effects of the Initial Population,Crossover and Mutation Rate to the Results of Genetic Algorithms and a Possible Solution Scheme
Shi Tang · Bulletin of Science and Technology · 2001
Our studies show that under the same genetic operators,effects of parameters,such as initial population,crossover rate and mutation rate,can't be ignored. Randomly generated initial population is not always suitable, and it's sometimes responsible for the unstable solutions. Unsuitable crossover and mutation rate can cause the same problem.For these reasons,an improved genetic algorism is presented in this paper.Self adjusting crossover and mutation rate are used and the initial population and population size are determined based on space division in this algorism.Results show that this method can avoid local convergence greatly.