An Improved Genetic Algorithm to Prevent Premature Convergence

An improved genetic algorithm is proposed to overcome premature convergence of the genetic algorithm.This improved algorithm uses a population diversity operator to initialize population with better distribution and to judge whether premature convergence occurs.Once premature convergence appears or tends to appear,the catastrophe operation is implemented to renew the population evolution of the algorithm.At the same time,a universal operator with selection and crossover operator is designed in combination with optimum individual and introduced random population in order to make the proposed algorithm's ability of maintaining population diversity and finding overall optimum solution.Experiments with four test functions demonstrate that the improved genetic algorithm can effectively maintain population diversity and prevent premature convergence.