A New Genetic Algorithm with Elliptical Crossover for Constrained Multi-objective Optimization Problems

The crossover operator plays an important role in a genetic algorithm, which produces two or more offspring for each pair of parents. With the help of the crossover operator, the genetic algorithm can explore the search space effectively. In this paper, we propose a new crossover operator called elliptical crossover operator, which can explore the search domain effectively. A local search scheme is designed to get more precise and wider nondominated solutions. In the local search scheme, the square search scheme and uniform design methods are combined. Based on the elliptical crossover operator and the local search scheme, a novel genetic algorithm is designed for constrained multi-objective optimization problems. Simulation results on several test functions indicates the effectiveness of the designed algorithm.