K-center Algorithm Based On Improved Glowworm Swarm Optimization

Data mining is a research hotspot in the domain of artificial intelligence and database.As an effective method of data mining,clustering analysis is to judge the correlation between data information through certain similarity criteria,and then cluster or cluster data objects.K-mediods is a classical clustering method based on partition,k-mediods selects the object closest to the mean center to represent the center of the class or cluster,which can effectively eliminate the impact of outliers on clustering effect.But it also has two defects,that is,it depends on the selection of initial cluster center and is easy to produce local optimal solution.Aiming at the defects of traditional k-mediods,this thesis proposes a new k-mediods algorithm based on improved glowworm swarm optimization.Firstly,the cloud model optimization strategy based on excellent glowworm swarm is combined with cloud dynamic adjustment factor strategy based on ordinary glowworm swarm and autonomous random search to improve the basic glowworm swarm optimization algorithm.The experimental results show that the improved GSO algorithm is superior to the basic GSO algorithm in the number of iterations,solution accuracy,solution space range and high-dimensional function.Secondly,according to the random convergence criterion,it is proved that the improved GSO converges to the global optimal solution with probability 1.Finally,in order to solve the defect that k-mediods depends on the initial cluster center,this paper combines the improved GSO algorithm with k-mediods.The purpose is to make use of the global search ability of the improved GSO,so that k-mediods can quickly and accurately find the best cluster center.Experiments on five UCI datasets show that the improved k-mediods algorithm has a certain improvement in convergence speed and clustering effect,and the convergence time of the algorithm is far less than that of k-medoids.In order to verify the feasibility of the improved k-mediods algorithm proposed in this thesis,the improved clustering algorithm is applied to the residential user tiered electricity price system.Based on the clustering analysis of the collected annual consumption data of household users,a tiered electricity price classification scheme suitable for local residents is worked out on the basis of the optimal clustering results.The experimental results show that the algorithm has a good application prospect in the tiered price optimization scheme.