Clustering Research Based On Fuzzy Algorithm And Radial Basis Neural Network

Cluster analysis has been widely used in various fields as an important data processing technology.In this paper,the fuzzy C and radial basis neural networks are used to study the clustering problem.This research provides a strategic reference for integrating the data information of things and exploring the laws of things.The main research contents and results of this paper can be summarized as follows:1.In the process of fuzzy clustering,considering the selection of the initial center point of clustering is sensitive to the whole clustering result,an initial clustering center selection method based on gray correlation degree is proposed to determine the initial clustering center.Make the clustering results more stable.2.In the clustering process,it is necessary to select the appropriate metric to evaluate the similarity in the data.Based on the classical Euclidean distance function as a measure of similarity,combined with the existing research results,the DTW distance and the SPDTW distance are introduced.Combining the advantages of the three distance metrics,the data may be dependent on a certain data.Distance or different degrees of dependence on different distance metrics are proposed.The three distance weighted forms are proposed as new similarity measures,and then applied to Fuzzy C-Medoids and Fuzzy C-Means hybrid fuzzy clustering techniques.3.Based on the weighted distance as the similarity measure,in order to better determine the weighted distance weight,we use the particle swarm optimization algorithm to optimize the weight,thus obtaining the global optimal solution and obtaining the optimal clustering result..The numerical experiments were carried out by MATLAB programming.The clustering analysis was carried out on different UCR datasets and SCADI standard datasets.The numerical results show that the clustering results of the improved fuzzy clustering algorithm are significantly improved.4.Combining unsupervised fuzzy clustering model and supervised learning neural network model,constructing an improved radial basis neural network clustering model.The method of selecting initial network training data based on fuzzy clustering is proposed.In the network training process,the improved fuzzy clustering algorithm is used to select the center of the radial basis function of the hidden center layer.Finally,by using different The UCI dataset and air measured data were numerically tested on this model,and the effectiveness of the improved neural network clustering was verified by comparison of clustering results.