Particle swarm optimization is a based-population heuristic global optimization technology and is referred to as a swarm-intelligence technique. In general, each particle is initialized randomly which increases the iteration time and makes the result unstable. In this paper, an improved clustering algorithm combined with Entropy-based Fuzzy Clustering (EFC) is presented. Firstly EFC algorithm gets an initial cluster center. Then the cluster center is regarded as inputs of one of all particles instead of being initialized randomly.