A Novel Genetic Algorithm Based on Individual and Gene Diversity Maintaining and Its Simulation

In order to overcome premature convergence in SGA, a novel adaptive genetic algorithm based on diversity maintaining is proposed. First, variance of all individuals´ fitness is used to measure individual diversity in a population and to adjust crossover probability adaptively. Second, to restrain the lack of effective genes in certain loci, mutation probabilities of all alleles in each locus vary adaptively depending on gene diversity in corresponding locus. We compare the performance of the DMAGA with that of the simple genetic algorithm (SGA) and AGA in optimizing several complex functions. The simulation result shows that the novel GA can obtain higher precision solution and avoid local optima.