Fuzzy clustering algorithm of interactive multi-sensor probabilistic data

Abstract

In traditional clustering algorithm, the number of classes must be set beforehand and it is difficult in setting parameters. For uncertain environment, the precision of clustering is low and the scalability is poor. To address these problems, a new fuzzy clustering algorithm for interactive multi-sensor probabilistic data is proposed in this paper. The optimal hierarchical fusion algorithm with no prior knowledge is used to sort the sensors used for fusion according to the quality and the importance of information. The fusion of the first layer is the fusion of probabilistic data of two interactive sensors. The fusion of the second layer is the fusion of the fusion results of the first layer and the probability data of the other sensor to obtain the final fusion results. On this basis, the fuzzy C mean clustering algorithm is proposed to cluster the interactive multi-sensor probabilistic data. Wireless sensor networks are dynamic, and it is difficult to determine the number of classes beforehand. Subtraction clustering algorithm is used to adaptively determine the number of classes and the initial cluster center though building mountain function as the data density index. Thus, the convergence speed of the algorithm is accelerated and the local optimum is avoided. Experimental results show that the proposed algorithm has high clustering accuracy and good scalability.