A New Strategy for Improving Particle Swarm Optimization

Particle swarm optimization (PSO) has proved its ability in solving complex search and optimization problems. From the earliest presentation of the algorithm, it has been acknowledged that the technique´s major weakness is its propensity to converge prematurely on early, possibly suboptimal solutions. In this paper, we propose some new strategies to improve the search performance of standard PSO. In order to balance the global search and local search ability, the new version of PSO adopts nonlinear decay approach to adjust the inertia weight and asynchronous time-varying approach to adapt the learning factors. Meanwhile, ldquofunction stretchrdquo technology is used to improve the local search performance. Two benchmark functions and a nonlinear constrained optimization problem are used to test the proposed algorithm. Experimental results show that the PSO with proposed modified strategies is effective and efficient.