A Novel Ecological Competitive Genetic Algorithm

Premature convergence is a well known problem that occurs with Genetic Algorithm (GA). Inspired by ecological competitive which can decrease survival and reproduction as the similar individuals approaches the carrying capacity, a novel Competitive Genetic Algorithm (CGA) is proposed to avoid the premature convergence of GA. Using the mechanism of competitive, the number of individuals who have the similar chromosome is kept in a reasonable lever, and the diversity of population is maintained. Comparing to other GAs, CGA doesn’t intervene three classical operations of GA (i.e. selection, crossover and mutation), it does competition according to the concentration when a new individual is born, and evolves a more diversiform and fitter generation. The algorithm of CGA is described in detail firstly, then analyzing the diversity of CGA, and an experiment is done. The results of the experiment reveal that both the distribution and the fitness of CGA are better then other GA’s.