Peter W. Tse

Wavelet Analysis and Envelope Detection For Rolling Element Bearing Fault Diagnosis—Their Effectiveness and Flexibilities

The components which often fail in a rolling element bearing are the outer-race, the inner-race, the rollers, and the cage. Such failures generate a series of impact vibrations in short time intervals, which occur at Bearing Characteristic Frequencies (BCF). Since BCF contain very little energy, and are usually overwhelmed by noise and higher levels of macro-structural vibrations, they are difficult to find in their frequency spectra when using the common technique of Fast Fourier Transforms (FFT). Therefore, Envelope Detection (ED) is always used with FFT to identify faults occurring at the BCF. However, the computation of ED is complicated, and requires expensive equipment and experienced operators to process. This, coupled with the incapacity of FFT to detect nonstationary signals, makes wavelet analysis a popular alternative for machine fault diagnosis. Wavelet analysis provides multi-resolution in time-frequency distribution for easier detection of abnormal vibration signals. From the results of extensive experiments performed in a series of motor-pump driven systems, the methods of wavelet analysis and FFT with ED are proven to be efficient in detecting some types of bearing faults. Since wavelet analysis can detect both periodic and nonperiodic signals, it allows the machine operator to more easily detect the remaining types of bearing faults which are impossible by the method of FFT with ED. Hence, wavelet analysis is a better fault diagnostic tool for the practice in maintenance.

Application of mother wavelet functions for automatic gear and bearing fault diagnosis

This paper introduces an automatic feature extraction system for gear and bearing fault diagnosis using wavelet-based signal processing. Vibration signals recorded from two experimental set-ups were processed for gears and bearing conditions. Four statistical features were selected: standard deviation, variance, kurtosis, and fourth central moment of continuous wavelet coefficients of synchronized vibration signals (CWC-SVS). In this research, the mother wavelet selection is broadly discussed. 324 mother wavelet candidates were studied, and results show that Daubechies 44 (db44) has the most similar shape across both gear and bearing vibration signals. Next, an automatic feature extraction algorithm is introduced for gear and bearing defects. It also shows that the fourth central moment of CWC-SVS is a proper feature for both bearing and gear failure diagnosis. Standard deviation and variance of CWC-SVS demonstrated more appropriate outcome for bearings than gears. Kurtosis of CWC-SVS illustrated the acceptable performance for gears only. Results also show that although db44 is the most similar mother wavelet function across the vibration signals, it is not the proper function for all wavelet-based processing.

Anomaly Detection Through a Bayesian Support Vector Machine

This paper investigates the use of a one-class support vector machine algorithm to detect the onset of system anomalies, and trend output classification probabilities, as a way to monitor the health of a system. In the absence of “unhealthy” (negative class) information, a marginal kernel density estimate of the “healthy” (positive class) distribution is used to construct an estimate of the negative class. The output of the one-class support vector classifier is calibrated to posterior probabilities by fitting a logistic distribution to the support vector predictor model in an effort to manage false alarms.

Classification of gear faults using cumulants and the radial basis function network

Every tooth in a gearbox is alternately meshing and detaching during its operation. Hence, the loading condition of the tooth is alternately changing. Such a condition will make the tooth easily subject to spalling and worn. Moreover, Gaussian type of noise which is always embedded in the measurements makes the signal-to-noise ratio (SNR) of the collected data low and difficult to extract in fault-related features. This paper aims to propose an approach for gear fault classification by using cumulants and the radial basis function (BRF) network. The use of cumulants can minimize Gaussian noise and increase the SNR. The RBF network has proven to be superior to back-propagation networks. The RBF network provides better functions to approximate non-linear inputs and faster in convergence. In this paper, experiments have been conducted on a real gearbox. The cumulants calculated from the vibration signal collected from the inspected gearbox are used as input features. The RBF network is then used as a classifier for various kinds of operating conditions of the gearbox. Results show that the method of classification by combining cumulants and the RBF network is promising and achieved better accuracy.

Development of an advanced noise reduction method for vibration analysis based on singular value decomposition

The paper developed a reasonable and practical method for identifying the useful information from the signal that has been contaminated by noise, so that to provide a feasible tool for vibration analysis. A new concept namely the Singular Entropy (SE) was proposed based on the singular value decomposition technique. With the aid of the SE, a series of investigations were done for discovering the distribution characteristics of noise contaminated and pure signals, and consequently an advanced noise reduction method was developed. The experiments showed that the proposed method was not only applied for dealing with the stationary signals but also applied for dealing with the non-stationary signals.

A comprehensive reliability allocation method for design of CNC lathes

Abstract In the design and development of computerized numerical control lathes, an effective reliability allocation method is needed to allocate system level reliability requirements into subsystem and component levels. During the allocation process, many factors have to be considered. Some of these factors can be measured quantitatively while others have to be assessed qualitatively. In this paper, we consider seven criteria for conducting reliability allocation. A comprehensive failure rate allocation method is proposed for conducting the task of reliability allocation. Example data from field studies are used to illustrate the proposed method.

Support vector data description for fusion of multiple health indicators for enhancing gearbox fault diagnosis and prognosis

A novel method for enhancing gearbox fault diagnosis and prognosis is developed by fusion of multiple health indicators through support vector data description. First, the Comblet transform is used to identify gear residual error signals from the raw signal. Second, based on the observation of gear residual error signals, a total of 11 gear health indicators are identified, and are categorized into two types of indicators. The first and second types of indicators are for fault diagnosis and prognosis, respectively. The first type has six indicators, which are sensitive to impulsive signals triggered by anomalous impacts. The second type has five indicators, which are suitable for tracking degradation of faults. Third, through the support vector data description, the first six health indicators are fused into type one indicators for fault diagnosis. The remaining five indicators are fused into type two indicators for fault prognosis. Finally, a Gaussian kernel is designed to enhance the performance of type one and two indicators by optimal range of width size. The effectiveness of the proposed method is validated through experiments. The new method has been proven to be superior to methods that use unfused indicators individually.

A fast and adaptive varying-scale morphological analysis method for rolling element bearing fault diagnosis

The research in fault diagnosis for rolling element bearings has been attracting great interest in recent years. This is because bearings are frequently failed and the consequence could cause unexpected breakdown of machines. When a fault is occurring in a bearing, periodic impulses can be revealed in its generated vibration frequency spectrum. Different types of bearing faults will lead to impulses appearing at different periodic intervals. In order to extract the periodic impulses effectively, numerous techniques have been developed to reveal bearing fault characteristic frequencies. In this study, an adaptive varying-scale morphological analysis in time domain is proposed. This analysis can be applied to one-dimensional signal by defining different lengths of the structure elements based on the local peaks of the impulses. The analysis has been first validated by simulated impulses, and then by real bearing vibration signals embedded with faulty impulses caused by an inner race defect and an outer race defect. The results indicate that by using the proposed adaptive varying-scale morphological analysis, the cause of bearing defect could be accurately identified even the faulty impulses were partially covered by noise. Moreover, compared to other existing methods, the analysis can be functioned as an efficient faulty features extractor and performed in a very fast manner.

Blind Source Separation and Blind Equalization Algorithms for Mechanical Signal Separation and Identification

Many advanced techniques have been developed for diagnosis of machine faults caused by vibration. They are effective if the inspected vibration is well isolated from interference caused by vibrations from adjacent components. However, the components of manufacturing machines are numerous, small, and packed closely together. Thus the signal collected by a sensor is the aggregate of vibrations from all nearby components. This, coupled with noise, makes it nearly impossible to detect the anomalous vibration generated by a particular component, especially those generated by small defective components. Recently, new signal processing methods, such as blind source separation (BSS) and blind equalization (BE), have been proposed to separate or recover the aggregated vibrations so that each source of vibration can be correctly identified. In this paper, a comparison study is presented. Some widely used BSS and BE algorithms have been compared to evaluate their performance in the separation of mechanical vibrations. Both simulated signals and real vibrations generated by industrial machines were used to verify the effectiveness of BSS and BE. Their deficiencies have also been identified and improvements are proposed in the paper, so that they could be effectively applied in the fault diagnosis of complex manufacturing machines.

Ensemble-approaches for clustering health status of oil sand pumps

Centrifugal slurry pumps are widely used in the oil sand industry, mining, ore processing, waste treatment, cement production, and other industries to move mixtures of solids and liquids. Wear of slurry pump components, caused by abrasive and erosive solid particles, is one of the main causes of reduction in the efficiency and useful life of these pumps. This leads to unscheduled outages that cost companies millions of dollars each year. Traditional maintenance strategies can be applied, but they provide insufficient warning of impending failures. On the other hand, condition monitoring and on-line assessment of the wear status of wetted components in slurry pumps are expected to improve maintenance management and generate significant cost savings for pump operators. In this context, the objective of the present work is to develop and compare two unsupervised clustering ensemble methods, i.e., fuzzy C-means and hierarchical trees, for the assessment and measurement of the wear status of slurry pumps when available data is extremely limited. The idea is to combine predictions of multiple classifiers to reduce the variance of the results so that they are less dependent on the specifics of a single classifier. This will also reduce the variance of the bias, because a combination of multiple classifiers may learn a more expressive concept class than a single classifier.