关键词: 大数据,聚类,粒子群,干扰抑制

Abstract: The data clustering method of large data sets has important research significance in pattern recognition,fault diagnosis,data mining and so on.Traditional data clustering algorithms use hybrid differential evolution particle swarm optimization algorithm,but because the cross term interference caused by interactions between components of data flow has an impact on the correct judgement,the clustering effect is not good.A large data clustering algorithm was proposed based on the time-frequency clustering algorithm.In the large database of Internet of things from the perspective of communication,big data clustering feature vector is generated.Arbitrary two cluster vectors are trained of membership grade about neighbor sample,and information is scheduled in time sliding window model.High frequency component suppression method is used to conduct interference suppression of time-frequency clustering interactions.By similarity fusion processing frequency domain convolution and using particle swarm optimization algorithm to do cluster fitness calculation,we improved the data clustering algorithm.The simulation results show that the algorithm has good anti-interference and self-adaptive performance,and it has high accuracy.

Key words: Large data,Clustering,Particle swarm,Interference suppression