Key Engineering Materials Vols. 413-414

Abstract: A new bearing fault recognition method based on volterra series and HMM is proposed. In the proposed method, first the feature vectors are extracted from amplitude demodulated signals obtained from normal, ball, inner and outer faulty bearings. The feature vectors are based on the volterra series of the vibration signals, which is obtained by the subspace method. Then these feature vectors are input to each fault’s HMM to be recognized. The result of experiment shows that the proposed method is very effective. The proposed method is tested with the experiment data sampled from drive end ball bearing of an induction motor driven mechanical system.

Abstract: In this paper, a novel similarity classifier which synthesizes the adaptive resonance theory (ART) and the similarity classifier based on the Yu’s norm is proposed. The proposed ART-similarity classifier can not only carry out training without forgetting previously trained patterns but also be adaptive to changes in the environment. In order to test the proposed classifier, it is applied to the fault diagnosis of rolling element bearings. Before application to the fault diagnosis of bearings, considering computation burden principal component analysis (PCA) is proposed to reduce the number of features. The PCs are input the proposed classifier to diagnose the faulty bearings. The experiment results testify that the proposed classifier can identify the faults accurately. Furthermore, in order to validate the effectiveness of the proposed classifier further, it compares with other neural networks, such as the fuzzy ART, self-organising feature maps (SOFMs) and radial basis function (RBF) neural network through diagnosing the bearings under the same conditions. The comparison results confirm the superiority of the proposed method.

Abstract: This paper introduces a quantitative measure based on the energy-to-Shannon entropy ratio for base wavelet selection in vibration signal analysis. The Gaussian-modulated sinusoidal signal and a realistic vibration signal measured from a ball bearing have been used to evaluate the effectiveness of the measure. Experimental results demonstrate that the wavelet selected using the developed measure is effective in diagnosing structural defects in the bearing and the method developed provides systematic guidance in wavelet selection.

Abstract: In order to monitor nonlinear production process effectively, multivariate statistical process control based on kernel principal component analysis is applied to process monitoring and diagnosis. Squared prediction error (SPE) statistic of the kernel principal component analysis (KPCA) model is used for process monitoring, and the fault causes of the production process could be tracked by the methods of data reconstruction and the optimal neighbor selection strategy. Simulation data and Tennessee Eastman process data are used for model validation, as a result the proposed method has better performance on abnormality detecting, compared with multivariate statistical process control based on linear principal component analysis. What is more, the causes of the faults are tracked effectively, thus the production process can be adjusted to prevent substandard products.

Abstract: In order to find the cause of the shaft crack appeared in a 600MW steam turbine generator, several measurements are taken. The shaft assembly consists of three parts: shrink-fitted coupling, shaft and key which are used to transfer torque between coupling and shaft. First, the natural frequency of the shaft system is calculated. The result shows that the vibration signal contains the frequency nears the second natural frequency of the steam turbine shaft system. Second, stress field near the damaged coupling and shaft is analyzed using finite element method. The torque under different operation condition is calculated before stress analysis and applied to the shaft assembly, then. And, local stress concentrations position calculated is proved to be coincident with the place crack initiation. Finally, the cause of the fault is analyzed: at least three causes have found to be having something to do with the crack fault.

Abstract: An experimental setup of rotor-bearing system is installed and vibration characteristics of the system with pedestal looseness are investigated. The pretightening bolt between the bearing house and pedestal is adjusted to simulate the pedestal looseness fault. The vibration waveforms, spectra and orbits are used to analyze the nonlinear response of the system with pedestal looseness. Different parameters, including speed, looseness gap, imbalance mass and disk position are changed to observe the nonlinear vibration characteristics. The experiments show that the system motion generally contains the 1/2X fractional harmonic component and multiple harmonic components such as 2X, 3X, etc. Under some special conditions, the pedestal looseness occurs intermittently, that is, occurs in some periods and doesn’t in other periods.

Abstract: Based on pattern spectrum entropy and proximal support vector machine (PSVM), a motor rolling bearing fault diagnosis method is proposed in this paper. It is very difficult to filter the fault vibration signals from the strong noise background because the roller bearing fault diagnosis is a problem of multi-class classification of inner ring fault, outer ring fault and ball fault. Firstly, vibration signals are processed by the pattern spectrum. Secondly, the morphological pattern spectrum entropy, and pattern spectrum values are utilized to identify the fault features of input parameters of PSVM classifiers. The experiment results demonstrate that the pattern spectrum quantifies various aspects of the shape-size content of a signal, and PSVM costs a little time and has better efficiency than the standard SVM.

Abstract: Fault diagnosis is a key technology to ensure safe and reliable operation of large electromechanical equipments. Fault feature extraction is important in fault diagnosis process. For non-stationary vibration signal, energy normalization method based on wavelet packet is used to make signal processing with signals decomposed into different frequency bands to perform energy normalization process in corresponding frequency band so as to extract fault feature. Flue Gas Turbine is chosen as research object and analysis shows that energy normalization processing method based on wavelet packet can effectively extract non-stationary signal characteristic parameters and make effective nondestructive testing analysis of equipments condition.

Abstract: The cyclic autocorrelation function is used with regard to the cyclostationarity of gear vibrations in order to extract the modulation features of gearbox vibration signals, and to detect localized gear damage. The properties of the amplitude and frequency modulated signals in the cyclic frequency domain are summarized in order to investigate the differences between the modulation features of normal and faulty gearbox vibration signals. Gear tooth spalling is detected by the presence of many sidebands in a zero-lag time-slice of the cyclic autocorrelation function, thereby indicating an increase in the degree of modulation effect. The damage source is located by the spacing of the sidebands.

Abstract: In the present study, a nonlinear system identification approach known as NARMAX (Nonlinear Auto-Regressive Moving Average with eXogenous Inputs) modelling method and the NOFRF (Nonlinear Output Frequency Response Function) are introduced to detect damage in plate. A set of NOFRF-based damage features is proposed, and the procedure about how to extract the features from the measured response data is presented in detail. An experimental application to the detection of damages in aluminium plates demonstrates the effectiveness and engineering significance of the new damage detection technique.