Particle swarm optimization clustering algorithm with mutation based on K-means

To deal with the K-means algorithm's defects of sensitivity to the initial cluster center and the premature convergence of particle swarm optimization algorithm,a particle swarm optimization clustering algorithm with mutation based on K-means is proposed.This algorithm compensates the shortcoming of K-means algorithm by using particle swarm optimization algorithm,determines the timing of K-means operation,according to the convergence of particles,and therefore improves search performance,and jumps out of local minima by using mutation operation.The K-means algorithm,PSO-K-means algorithm and the algorithm proposed above are used to test the clustering of the three kinds of actual data.The comparison of the experimental results show that the algorithm can jump out of local minima,and it is able to find a better solution than the other two algorithms,therefore more efficient and more stable.