Vector Spectrum—Fuzzy Clustering And The Research Of Its Application In Fault Diagnosis

With the advancement of modern science and technology, the equipments tend to be larger, faster as well as more roboticized and complicated, which makes technology of state monitoring and fault diagnosis more and more important. Based on homologous fusion Vector spectrum analysis technique can reflect a more comprehensive and accurate information about spatial characteristics of the rotor movement. Besides it can be compatible with traditional spectrum analysis. At the same time the numerical algorithm is simple, robust, and fast so that it is more conducive to intelligent fault diagnosis. As a kind of pattern recognition, the fuzzy clustering method’s most classic algorithm is fuzzy C-means algorithms. Because of the sensitivity of the structure of its data set and the initial value, it is difficult to obtain the optimal clustering. Therefore, this paper proposes two improved algorithms. Based on that, this paper will combine the full vector spectrum technology with improved FCM algorithm, and then apply to fault diagnosis of rotating machinery. The main work is done as follows:1. Vector spectrum which based on congenetic information fusion are expounded in this thesis, and its fast and steadiness numeric algorithms be elicited. And it proved the advantage of vector spectrum compare to the traditional methods by experiment results. Then this paper proposes the feature extraction method which is based on the vector spectrum and lay the groundwork for the accurate classification of fuzzy clustering.2. Introduction of the density function and the kernel function method in the classic FCM, in this paper, the fuzzy kernel function algorithm based on density function (DKFCM) was proposed. The algorithm’s specific steps was given Combine Vector Spectrum and DKFCM algorithm, and applied it to rotating machinery fault diagnosis, then the specific fault classification process was given. The experimental results showed the validity and the advantage of this new method:the classification accuracy higher and the iterative times less than the traditional method. 3. The Genetic Algorithm (GA) and the Uniform Design method were introduced into the classical FCM algorithm, and the new methods-Uniform Genetic Fuzzy Clustering Algorithm (UGAFCM) was presented. Do feature extraction by Vector Spectrum technology, build a classifier using UGAFCM algorithm, establishes the mathematical model of fault diagnosis method and give the specific diagnosis flowchart. At last, the experiment research was done, and the result shows that this proposed approach is effective.4. In view of the fault diagnosis, fault number is always difficult to determine in advance, therefore this paper introduces the fuzzy cluster validity indices. Because the effect of the cluster validity index has domain correlation, the effective indices which are applicable to the field of rotating machinery fault diagnosis are preliminary discussed in this paper.