Abstract: In trying to solve constrained optimization problems by particle swarm optimization, the way to handle the constrained conditions is th e key factor for success. Some features of particle swarm optimization and a lar ge number of constrained optimization problems are taken into account and then a new method is proposed, which means to separate the objective functions from it s constrained functions. Therefore, every particle of particle swarm optimiz ation has double fitness values whether the particle is better or not will be de cided by its two fitness values. The strategy to keep a fixed proportion of infe asible individuals is used in this new method. Numerical results show that t he improved PSO is feasible and can get more precise results than particle swarm optimization by using penalty functions and genetic algorithm and other optimiz ation algorithms.

Key words: particle swarm optimization, double fitness value, adaptive