Research On Intelligent Fault Diagnosis Of Rotating Machinery Based On CEEMD And Texture Feature

Intelligent fault diagnosis of rotating machinery based on data doesn't needs prior knowledge and experience of experts, this method has stronger applicability than Traditional method of equipment fault diagnosis, and it is the main development direction in the field of equipment fault diagnosis research. Vibration signal denoising method damages integrity of signal, fault feature sensitivity is poor and the pattern classifier parameters optimization algorithm convergence speed is slow. In view of this problems in intelligent fault diagnosis method of rotating machinery, this paper proposed a number of corresponding key technology improvements to enhance the ability and precision of the intelligent fault diagnosis method of rotating machinery. The specific contents are as follows:First of all, study on reducing the environmental noise of the vibration signal of rotating machinery is necessary. The vast majority of rotating machinery are in complex working environment, so it needs effective noise reduction as the same time avoid signal information damage. In order to achieve the purpose of signal processing, a relatively new method of signal denoising is formed by the combination of CEEMD and wavelet threshold denoising method, which can complete noise reduction and ensure the signal integrity as the same time.Secondly, this paper proposed a new fault feature, SPWVD image is extracted from the vibration signal after noise reduction, and the texture feature parameters of the image are extracted from the time-frequency distribution image to form the new feature.Using the new fault feature to achieve better information mining in each stage of fault signal, in order to further improve the ability and accuracy of rotating machinery fault diagnosis process. Then the performance of the new feature was verified by experiment and comparison of several different time-frequency analysis method to extract fault features.Again, this paper establish a rotating machinery intelligent fault pattern classifier based on SVM, a improved adaptive particle swarm optimization algorithm is used to optimize the parameters of the pattern classifier to improve the accuracy of the intelligent diagnosis process. Training SVM with the training fault feature to get the best pattern classifier, then use the test fault feature as input to get pattern recognition results, achieving the intelligent fault diagnosis of rotating machinery.Finally, the reliability was validated by an actual example. The vibration data which is obtained in the actual rotating machinery fault diagnosis experiment is used as input to make four sets of simulation experiments. The goal of these experiments is to identify the fault status of the rotating machinery and the different faults degree in one fault condition of the rotating machinery, then the method is simulated and verified in the practical application.This paper provides a kind of intelligent fault diagnosis of rotating machinery. It improved vibration signal denoising method, fault feature extraction method and fault pattern classifier parameter optimization method. This method has important significance for future research of intelligent fault diagnosis of rotating machinery.