A Clustering K-means Algorithm Based on Improved PSO Algorithm

Because of the shortcomings of the traditional K-means algorithm which is sensitive to select the initial clustering centers and easy converges to local optimization, the paper proposes An Clustering K-means Algorithm Based on Improved the Particle Swarm Optimization Algorithm. The algorithm uses a powerful global search capability of the Particle Swarm Optimization algorithm to optimize the selection of the initial clustering centers: dynamically adjusting the inertia weight and other parameters to enhance the performance of the Particle Swarm Optimization; taking advantage of the fitness variance of the group to decide the conversion timing between the front part of the Particle Swarm Optimization algorithm and the rear part of K-means algorithm; setting the variables to monitor the changes of the optimal values of each particle and particle population, timely take the premature convergence particle to the mutation operation, thus we can find the global optimum initial clustering centers for K-means algorithm, then the clustering results are not affected by the initial clustering centers, it is easy to get global optimal solution. The experimental results show that the clustering accuracy rate, clustering quality and the global search capabilities of the improved algorithm is higher than the traditional clustering algorithm proposed in this paper.