Diagnostics Of Rolling Element Bearings Research Articles

A Comparison of Signal Analysis Techniques for the Diagnostics of the IMS Rolling Element Bearing Dataset

In this paper, a comparison of signal analysis techniques for the diagnostics of rolling element bearings is carried out. Specifically, the comparison is performed in terms of fault detection, diagnosis and prognosis techniques with regards to the first rolling element bearing dataset released by NASA IMS Center in 2014. As for fault detection, it is obtained that RMS value, Kurtosis and Detectivity, as statistical parameters, are able to properly detect the arising of the fault on the defective bearings. Then, several signal processing techniques, such as deterministic/random signal separation, time-frequency and cyclostationary analyses are applied to perform fault diagnosis. Among these techniques, it is found that the combination of Cepstrum Pre-Whitening and Squared Envelope Spectrum, and Improved Envelope Spectrum, allow the faults to be correctly identified on specific bearing components. Finally, the Correlation, Monotonicity and Robustness of the previous statistical parameters are computed to identify the most accurate tools for bearing fault prognosis.

Why EMD and similar decompositions are of little benefit for bearing diagnostics

Empirical mode decomposition (EMD) is a way of decomposing complex signals into a sum of “mono-components”, i.e. intrinsic mode functions (IMFs), each of which can be considered as a single carrier frequency, modulated in both amplitude and phase/frequency (an analytic signal), so that both amplitude and phase remain continuous and differentiable, the latter to instantaneous frequency. In many cases, the instantaneous frequency and amplitude could be valuable diagnostic features, for example with gear vibrations. Many authors have suggested applying this technique to the diagnostics of rolling element bearings (REBs), but this paper shows that REB signals intrinsically do not have such properties, except perhaps in short bursts, so that the excessive computation time trying to ensure continuous phase is wasted. Much more appropriate methods exist to extract the optimum envelope signal, which contains the desired diagnostic information, based on more relevant criteria. As its name suggests, the EMD process is empirical, with no analytical solution, so that different results can be obtained from different segments of the same signal, partly because of different noise content. Another way of obtaining IMFs is by using variational mode decomposition (VMD), which does have an analytical solution, making it less sensitive to noise, but still assumes that the signal can be validly decomposed into mono-components, which is not the case for bearing signals. The EMD process has two additional problems which make difficult its use for REB signals; (1) end effects, which usually mean that many, perhaps a majority, of IMFs are required purely to compensate for them, and which cannot be truncated for signals consisting of short bursts with sections of noise between them; and (2) mode mixing, meaning that repeatable results cannot be guaranteed, even for valid mono-components, let alone REB signals which are of a stochastic nature. The paper explains in detail why REB signals cannot be validly decomposed by EMD and similar methods, to give sub-components with properties directly related to bearing faults, even though many existing papers describe how the IMFs found by the EMD process might be selected or re-combined to have properties of interest in bearing diagnostics. In particular, the paper shows that the equivalent bandpass filters given by EMD have much poorer selectivity than most alternative filters. The paper uses many simulated and measured examples and demonstrations, to discuss these claims, including a number based on results from already published papers.

Domain Knowledge Informed Unsupervised Fault Detection for Rolling Element Bearings

Early and accurate detection of rolling element bearing faults in rotating machinery is important for minimizing production downtime and reducing unnecessary preventative maintenance. Several fault detection methods based on signal processing and machine learning methods have been proposed. Particularly, supervised, data-driven approaches have proved to be very effective for fault detection and diagnostics of rolling element bearings. However, supervised methods rely heavily on the availability of failure data with volume, variety and veracity, which is mostly unavailable in industry. As an alternative data-driven strategy, unsupervised methods are trained on healthy data only and do not require any failure data. In contrast to supervised and un-supervised data-driven models, physics-based and phenomenological models are based on domain knowledge and not on historical data. Although these models are useful for studying the way in which damage is expected to manifest in a measured signal, they are difficult to calibrate and often lack the fidelity required to model reality. In this paper, an unsupervised data-driven anomaly detection method that exploits informative domain knowledge is proposed. Hereby, the versatility of unsupervised data-driven methods are combined with domain knowledge. In this approach, supplementary training data is generated by augmenting healthy data towards its possible future faulty state based on the characteristic bearing fault frequencies. Both healthy and augmented squared envelope spectrum data is used to train an autoencoder model that includes regularisation designed to constrain the latent features at the autoencoder bottleneck. Regularisation in the autoencoder loss enforces that the expected deviation of the healthy latent representation towards the augmented latent representation at dam aged conditions, is constrained to be maximally different for different fault modes. Consequently, the likelihood of a new test sample being healthy can be evaluated based on the projection of the sample onto an expected failure direction in the latent representation. A phenomenological and experimental dataset is used to demonstrate that the addition of augmented training data and a specialized autoencoder loss function can create a separable latent representation that can be used to generate interpretable health indicators.

Recent advancements of signal processing and artificial intelligence in the fault detection of rolling element bearings: a review

A rolling element bearing is a common component in household and industrial machines. Even a minor fault in this section has a negative impact on the machinery's overall operation. As a result, the industry suffers significant financial losses, and this damage can potentially result in catastrophic failures. Therefore, even a little fault in the rolling element bearings must be recognized and remedied as soon as possible. Many ways for detecting REB defects have been created in recent years, and new methods are being introduced on a daily basis. This article will provide a summary of such methods, with a focus on vibration analysis techniques. The newest advancements in this field will be recognizable to readers of this article. Anyone interested in defect diagnostics of rolling element bearings can utilize this material.

Deep semi-supervised generative adversarial fault diagnostics of rolling element bearings

With the availability of cheaper multisensor suites, one has access to massive and multidimensional datasets that can and should be used for fault diagnosis. However, from a time, resource, engineering, and computational perspective, it is often cost prohibitive to label all the data streaming into a database in the context of big machinery data, that is, massive multidimensional data. Therefore, this article proposes both a fully unsupervised and a semi-supervised deep learning enabled generative adversarial network-based methodology for fault diagnostics. Two public datasets of vibration data from rolling element bearings are used to evaluate the performance of the proposed methodology for fault diagnostics. The results indicate that the proposed methodology is a promising approach for both unsupervised and semi-supervised fault diagnostics.

Rolling Element Bearing Diagnostics Using Extended Phase Space Topology

This paper introduces a novel method called extended phase space topology (EPST) for machinery diagnostics and pattern recognition. In particular, the research focuses on fault detection and diagnostics of rolling element bearings. The proposed method is based on mapping the vibrational response onto the density space and approximating the density using orthogonal functions. The method has been applied to vibration data of a rotating machine where the data were measured by proximity probes. The method was applied to two operating conditions: constant operating speed and variable operating speed. As will be shown, the proposed feature extraction method has an outstanding capability in characterizing the system response and diagnosing the system. The method is evidently robust to noise, does not depend on expert knowledge about the system, requires no feature ranking or selection, and can easily be applied in an automated process. Finally, a comparison with utilization of statistical features is performed for each operating condition, which demonstrates that the proposed method performs better than the traditional statistical methods.

Application of cepstrum pre-whitening for the diagnosis of bearing faults under variable speed conditions

Diagnostics of rolling element bearings involves a combination of different techniques of signal enhancing and analysis. The most common procedure presents a first step of order tracking and synchronous averaging, able to remove the undesired components, synchronous with the shaft harmonics, from the signal, and a final step of envelope analysis to obtain the squared envelope spectrum. This indicator has been studied thoroughly, and statistically based criteria have been obtained, in order to identify damaged bearings. The statistical thresholds are valid only if all the deterministic components in the signal have been removed. Unfortunately, in various industrial applications, characterized by heterogeneous vibration sources, the first step of synchronous averaging is not sufficient to eliminate completely the deterministic components and an additional step of pre-whitening is needed before the envelope analysis. Different techniques have been proposed in the past with this aim: The most widely spread are linear prediction filters and spectral kurtosis. Recently, a new technique for pre-whitening has been proposed, based on cepstral analysis: the so-called cepstrum pre-whitening. Owing to its low computational requirements and its simplicity, it seems a good candidate to perform the intermediate pre-whitening step in an automatic damage recognition algorithm. In this paper, the effectiveness of the new technique will be tested on the data measured on a full-scale industrial bearing test-rig, able to reproduce the harsh conditions of operation. A benchmark comparison with the traditional pre-whitening techniques will be made, as a final step for the verification of the potentiality of the cepstrum pre-whitening.

A New Morphological Method for Faults Diagnosis of Rolling Element Bearings

A new morphology analysis method had been proposed to effectively extract the impulse components in the vibration signals of defective rolling element bearings. In the method, the morphology operator had been constructed by average of the closing and opening operator. For the construction of structure element (SE), the flat and zero was adopted as the shape and the height of SE, respectively, and the element numbers of the SE was optimized by a new proposed criterion (called SNR criterion). Vibration signals of two defective rolling bearings with an outer and an inner fault respectively are employed to validate the proposed method and the results are compared with ones calculated by envelopment analysis method. It shows that the proposed method is effective and robust to extract morphological features, and can be used to the on-line diagnostics of rolling element bearings in rotating machines conveniently.

Application of support vector machine based on pattern spectrum entropy in fault diagnostics of rolling element bearings

This paper presents a novel pattern classification approach for the fault diagnostics of rolling element bearings, which combines the morphological multi-scale analysis and the ‘one to others’ support vector machine (SVM) classifiers. The morphological pattern spectrum describes the shape characteristics of the inspected signal based on the morphological opening operation with multi-scale structuring elements. The pattern spectrum entropy and the barycenter scale location of the spectrum curve are extracted as the feature vectors presenting different faults of the bearing, which are more effective and representative than the kurtosis and the enveloping demodulation spectrum. The ‘one to others’ SVM algorithm is adopted to distinguish six kinds of fault signals which were measured in the experimental test rig under eight different working conditions. The recognition results of the SVM are ideal and more precise than those of the artificial neural network even though the training samples are few. The combination of the morphological pattern spectrum parameters and the ‘one to others’ multi-class SVM algorithm is suitable for the on-line automated fault diagnosis of the rolling element bearings. This application is promising and worth well exploiting.

Morphological undecimated wavelet decomposition for fault diagnostics of rolling element bearings

This paper presents a novel morphological undecimated wavelet (MUDW) decomposition scheme for fault diagnostics of rolling element bearings. The MUDW scheme is developed based on the morphological wavelet (MW) theory for both the extraction of impulse features and noise smoothing in signal processing. The analysis operators and the synthesis operator of MUDW strictly satisfy the pyramid condition. The MUDW scheme is used to extract impulse features from rolling element bearing defect signals imposed with noise. The efficiency of the MUDW scheme used for noise smoothing and the extraction of impulse components is evaluated using the simulated data and measured signals from the bearing test rig. Compared with enveloping demodulation analysis, the MW transform and the traditional wavelet transform (WT), the MUDW decomposition scheme is more effective and suitable for the on-line diagnostics of bearings in rotating machines.

The enhancement of fault detection and diagnosis in rolling element bearings using minimum entropy deconvolution combined with spectral kurtosis

Spectral kurtosis (SK) represents a valuable tool for extracting transients buried in noise, which makes it very powerful for the diagnostics of rolling element bearings. However, a high value of SK requires that the individual transients are separated, which in turn means that if their repetition rate is high their damping must be sufficiently high that each dies away before the appearance of the next. This paper presents an algorithm for enhancing the surveillance capability of SK by using the minimum entropy deconvolution (MED) technique. The MED technique effectively deconvolves the effect of the transmission path and clarifies the impulses, even where they are not separated in the original signal. The paper illustrates these issues by analysing signals taken from a high-speed test rig, which contained a bearing with a spalled inner race. The results show that the use of the MED technique dramatically sharpens the pulses originating from the impacts of the balls with the spall and increases the kurtosis values to a level that reflects the severity of the fault. Moreover, when the algorithm was tested on signals taken from a gearbox for a bearing with a spalled outer race, it shows that each of the impulses originating from the impacts is made up of two parts (corresponding to entry into and exit from the spall). This agrees well with the literature but is often difficult to observe without the use of the MED technique. The use of the MED along with SK analysis also greatly improves the results of envelope analysis for making a complete diagnosis of the fault and trending its progression.

A Stochastic Model for Simulation and Diagnostics of Rolling Element Bearings With Localized Faults

This paper addresses the stochastic modeling of the vibration signal produced by localized faults in rolling element bearings and its use for diagnostic purposes. The aim is essentially to provide a better understanding of the recognized “envelope analysis” technique as classically used in the diagnostics of rolling element bearings, and incidentally give theoretical proofs for the specific features of envelope spectra as obtained from experimental data. The proposed model may also prove useful for simulation purposes. First, the excitation force generated by a defect is modeled as a random point process and its spectral signature is derived analytically. Then its transmission through the bearing is investigated in detail in order to find the spectral characteristics of the resulting vibration signal. The analysis finally gives sound justification for “squared” envelope analysis and the type of spectral indicators that should be used with it.