Research On Rotating Machinery Vibration Fault Diagnosis Based On Support Vector Machine

Nowadays science and technology have developed rapidly. Large rotating machinery is developing toward large-scale, complex, intelligent direction. Condition monitoring and fault diagnosis of rotating machinery receive more and more attention. Intelligent fault diagnosis technology is also gradually developed. Support vector machine is a kind of based on statistical learning theory emerging theory of machine learning, it can effectively solve the problem of small sample fault classification. So SVM is introduced into the rotating machinery fault diagnosis technology field, it can diagnose fault conveniently and effectively. In order to avoid huge economic losses and serious casualties, the study of large rotating machinery intelligent fault diagnosis has great significance.This thesis mainly according to the theory of support vector machine algorithm as the research method, the intelligent fault diagnosis technology has carried on the detailed experimental studies. This paper studies the theory of algorithms and fault diagnosis pattern recognition SVM in detail, and designed with LabVIEW and Matlab software development platform of rotating machinery fault diagnosis system based on support vector machine. The main research content is as follows:This paper studied the large rotating machinery fault diagnosis technology application and development situation, detailed study of the statistical learning theory and support vector machine theory. The paper summarized the SVM current situation of development and SVM in the fault diagnosis technology research and development. For application of support vector machine in rotating machinery fault diagnosis research has laid a good foundation.With LabVIEW and Matlab as the software development platform, a set of intelligent system of rotating machinery vibration test and fault diagnosis is designed, the system include: data acquisition, data processing, signal analysis, data management, feature extraction and fault diagnosis module. System can be effective for data collection and processing, and make time-frequency analysis of vibration signal. In the system we use wavelet packet analysis method to decompose and reconstruct the collected vibration signal. We use the wavelet packet energy value obtained as the feature vector input into the support vector machine-based pattern recognition module. Establish SVM training model to achieve fault recognition of rotating machinery.Through large rotating machinery vibration data measured at the site to verify the fault diagnosis system of rotating machinery vibration based on SVM. We can optimize the parameters of SVM and improve the design of system through lots of experiments. Implementing the vibration fault classification based on SVM so as to achieve the aim of rotating machinery fault, early detection.