Abstract: Aiming at the constrained optimization probl ems, we introduced the concept of semi-feasible region, proposed a novel rule of tournament selection, and improved the fitness function of evolutionary algorithm which is based on tournament selection and penalty function. Making use of characteristics of Particle Swarm Optimization (PSO), we designed a selection operator for the semi-feasible region and proposed a novel evolutionary algorithm for solving constrained optimization problems. Numerical experiments demonstrate the effectiveness of the proposed algorithm.

Key words: constrained optimization problems, particle swarm optimization, feasible region, tournament selection