Search

Continuous long-term monitoring of motor health is crucial for the early detection of abnormalities such as bearing faults (up to 51% of motor failures are attributed to bearing faults). Despite numerous methodologies proposed for bearing fault detection, most of them require normal (healthy) and abnormal (faulty) data for training. Even with the recent deep learning (DL) methodologies trained on the labeled data from the same machine, the classification accuracy significantly deteriorates when one or few conditions are altered, e.g., a different speed or load, or for different fault types/severities with sensors placed in different locations. Furthermore, their performance suffers significantly or may entirely fail when they are tested on another machine with entirely different healthy and faulty signal patterns. To address this need, in this pilot study, we propose a zero-shot bearing fault detection method that can detect any fault on a new (target) machine regardless of the working conditions, sensor parameters, or fault characteristics. To accomplish this objective, a 1D Operational Generative Adversarial Network (Op-GAN) first characterizes the transition between normal and fault vibration signals of (a) source machine(s) under various conditions, sensor parameters, and fault types. Then for a target machine, the potential faulty signals can be generated, and over its actual healthy and synthesized faulty signals, a compact, and lightweight 1D Self-ONN fault detector can then be trained to detect the real faulty condition in real time whenever it occurs. To validate the proposed approach, a new benchmark dataset is created using two different motors working under different conditions and sensor locations. Experimental results demonstrate that this novel approach can accurately detect any bearing fault achieving an average recall rate of around 89% and 95% on two target machines regardless of its type, severity, and location. [ABSTRACT FROM AUTHOR]

Abstract Adaptive stochastic resonance (ASR) has been proven effective in enhancing weak periodic signals that are submerged in heavy background noise. Given such benefit, ARS has also been applied in detecting bearing faults based on vibration signal analysis. However, when the vibration has an extremely low signal-to-noise ratio (SNR), the fault characteristic frequency may not be accurately enhanced via the traditional ASR. To address this problem, this paper designs the sound-aided vibration signal ASR (SAVASR) method, which procedures are summarized as follows. First, the bearing sound and vibration signals are demodulated. Second, the envelope vibration signal is adaptively enhanced by moving a sliding window along the time axis of the envelope sound signal. Third, the optimized fused signal is sent to the ASR system, in which the parameters are adaptively adjusted based on a synthetic evaluation index. Fourth, the bearing fault is detected from the spectrum of the optimal SAVASR output signal. Qualitative and quantitative analyses are performed to evaluate and compare the performance of SAVASR with that of ASR, where only the vibration signal is processed. Given its unique approach in detecting weak signals by fusing multiple sensor information, SAVASR shows high potential in automatically detecting bearing faults especially under low SNR conditions. [ABSTRACT FROM AUTHOR]

Highlights • A very first real application of RS on vibration to evaluate its potential in machine monitoring. • A simulation study to evaluate the effect of RS parameters within its different modes. • An enhanced method for spectral analysis of RS signals based on zero insertion and Welch periodogram. Abstract Random sampling was first introduced in the 1960's (Beutler and Leneman, 1966 [1]), and was then reused in the Compressive Sensing theory (published in 2006 (Donoho, 2006 [2])) for having two important advantages: anti-aliasing property and low sampling rate. These advantages decrease the demand of high speed acquisition and big data storage. In this paper, a brief review of the principal random sampling modes is presented: Additive Random Sampling (ARS) and Jittered Random Sampling (JRS) that both use probability distributions: uniform or Gaussian (Wojtiuk, 2000; Ben Romdhane, 2009; Luo, 2012; Lo and Purvis, 1997 [3–6]). In fact, the stationarity of the random point process is the essential criterion to guarantee the anti-aliasing property (Bilinksis and Mikelson, 1992 [7]). Accordingly, these modes of sampling are studied and used in simulation and hardware implementation. In addition, a study of the time quantization effect is done to explore the different aspects of discrete random sampling (Ben Romdhane, 2009; Luo, 2012 [4,5]). The performance of this sampling is approved by simulation and practice, using respectively Matlab and Arduino microcontroller, respectively, for acquiring signals at a randomly spaced clock. A database was created in order to observe the impact of statistical parameters such as the mean and the standard deviation for Gaussian distribution, and the interval endpoints of uniform distribution. In conclusion, the ARS with uniform distribution has been proved to have less limitations (Ben Romdhane, 2009 [4]), whereas the spectrum of sampled signal is free of alias and the sampling frequency is reduced (a much smaller than Nyquist-Shannon sampling rate). In order to take advantage of this mode, it is applied on vibration signals to enhance remoted machine monitoring. [ABSTRACT FROM AUTHOR]

Envelope analysis is a widely used method for rolling element bearing fault detection. To obtain high detection accuracy, it is critical to determine an optimal frequency narrowband for the envelope demodulation. However, many of the schemes which are used for the narrowband selection, such as the Kurtogram, can produce poor detection results because they are sensitive to random noise and aperiodic impulses which normally occur in practical applications. To achieve the purposes of denoising and frequency band optimisation, this paper proposes a novel modulation signal bispectrum (MSB) based robust detector for bearing fault detection. Because of its inherent noise suppression capability, the MSB allows effective suppression of both stationary random noise and discrete aperiodic noise. The high magnitude features that result from the use of the MSB also enhance the modulation effects of a bearing fault and can be used to provide optimal frequency bands for fault detection. The Kurtogram is generally accepted as a powerful means of selecting the most appropriate frequency band for envelope analysis, and as such it has been used as the benchmark comparator for performance evaluation in this paper. Both simulated and experimental data analysis results show that the proposed method produces more accurate and robust detection results than Kurtogram based approaches for common bearing faults under a range of representative scenarios. [ABSTRACT FROM AUTHOR]

The order tracking method based on time-frequency representation is regarded as an effective tool for fault detection of bearings with varying rotating speeds. In the traditional order tracking methods, a tachometer is required to obtain the instantaneous speed which is hardly satisfied in practice due to the technical and economical limitations. Some tacholess order tracking methods have been developed in recent years. In these methods, the instantaneous frequency ridge extraction is one of the most important parts. However, the current ridge extraction methods are sensitive to noise and may easily get trapped in a local optimum. Due to the presence of noise and other unrelated components of the signal, bearing fault features are difficult to be detected from the envelope spectrum or envelope order spectrum. To overcome the abovementioned drawbacks, an adaptive and tacholess order analysis method is proposed in this paper. In this method, a novel ridge extraction algorithm based on dynamic path optimization is adopted to estimate the instantaneous frequency. This algorithm can overcome the shortcomings of the current ridge extraction algorithms. Meanwhile, the enhanced empirical wavelet transform (EEWT) algorithm is applied to extract the bearing fault features. Both simulated and experimental results demonstrate that the proposed method is robust to noise and effective for bearing fault detection under variable speed conditions. [ABSTRACT FROM AUTHOR]

It is a fundamental task in the machine fault diagnosis community to detect impulsive signatures generated by the localized faults of bearings. The main goal of this paper is to exploit the low-rank physical structure of periodic impulsive features and further establish a weighted low-rank sparse model for bearing fault detection. The proposed model mainly consists of three basic components: an adaptive partition window, a nuclear norm regularization and a weighted sequence. Firstly, due to the periodic repetition mechanism of impulsive feature, an adaptive partition window could be designed to transform the impulsive feature into a data matrix. The highlight of partition window is to accumulate all local feature information and align them. Then, all columns of the data matrix share similar waveforms and a core physical phenomenon arises, i.e., these singular values of the data matrix demonstrates a sparse distribution pattern. Therefore, a nuclear norm regularization is enforced to capture that sparse prior. However, the nuclear norm regularization treats all singular values equally and thus ignores one basic fact that larger singular values have more information volume of impulsive features and should be preserved as much as possible. Therefore, a weighted sequence with adaptively tuning weights inversely proportional to singular amplitude is adopted to guarantee the distribution consistence of large singular values. On the other hand, the proposed model is difficult to solve due to its non-convexity and thus a new algorithm is developed to search one satisfying stationary solution through alternatively implementing one proximal operator operation and least-square fitting. Moreover, the sensitivity analysis and selection principles of algorithmic parameters are comprehensively investigated through a set of numerical experiments, which shows that the proposed method is robust and only has a few adjustable parameters. Lastly, the proposed model is applied to the wind turbine (WT) bearing fault detection and its effectiveness is sufficiently verified. Compared with the current popular bearing fault diagnosis techniques, wavelet analysis and spectral kurtosis, our model achieves a higher diagnostic accuracy. [ABSTRACT FROM AUTHOR]

Vibration time signals emitted by a defective rolling element bearing are often used to determine its health condition. For real-world measurements these fault signals may be weak compared to the remaining content in the measured signal causing increased difficulties to fault detection. One way of enhancing the fault detection is by demodulation at a filtered frequency band in which the Signal-to-Noise-Ratio (SNR) of the fault signal is high. In the present paper, we propose a method that uses the resonances obtained from Operational Modal Analysis (OMA) as center frequencies for the filter bands as it is assumed that the characteristic bearing fault frequencies modulate resonances or carrier frequencies close to them. An indicator is used to determine which of the resonances from the OMA enhance the fault signal the most and this selection is dependent on the bandwidth as well. The presented method is applied on signals from two experimental cases, an experimental test rig of a planetary gearbox and a wind turbine gearbox, and is compared to the Fast Kurtogram and Log-Cycligram to assess its performance for band selection. The results show that the presented method determines a suited band for band pass filtering which enhances the fault detection at the two complex use cases. • Operational modal analysis (OMA) method for enhancing fault detection. • Fault detection of rolling element bearings by demodulation at resonances. • Center frequencies for band-pass filtering determined by OMA. • Indicator determines the optimal band. • Method evaluated on challenging real-world signals. [ABSTRACT FROM AUTHOR]

We report a non-resonance based approach to bearing fault detection. This is achieved by a higher order energy operator fusion (HOEO_F) method. In this method, multiple higher order energy operators are fused to form a single simple transform to process the bearing signal obscured by noise and vibration interferences. The fusion is guided by entropy minimization. Unlike the popular high frequency resonance technique, this method does not require the information of resonance excited by the bearing fault. The effects of the HOEO_F method on signal-to-noise ratio (SNR) and signal-to-interference ratio (SIR) are illustrated in this paper. The performance of the proposed method in handling noise and interferences has been examined using both simulated and experimental data. The results indicate that the HOEO_F method outperforms both the envelope method and the original energy operator method. [ABSTRACT FROM AUTHOR]

Condition monitoring of rotating machines in speed-varying conditions remains a challenging task and an active field of research. Specifically, the produced vibrations belong to a particular class of non-stationary signals called cyclo-non-stationary: although highly non-stationary, they contain hidden periodicities related to the shaft angle; the phenomenon of long term modulations is what makes them different from cyclostationary signals which are encountered under constant speed regimes. In this paper, it is shown that the optimal way of describing cyclo-non-stationary signals is jointly in the time and the angular domains. While the first domain describes the waveform characteristics related to the system dynamics, the second one reveals existing periodicities linked to the system kinematics. Therefore, a specific class of signals – coined angle-time cyclostationary is considered, expressing the angle-time interaction. Accordingly, the related spectral representations, the order-frequency spectral correlation and coherence functions are proposed and their efficiency is demonstrated on two industrial cases. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, an original method for bearing fault detection in high speed synchronous machines is presented. This method is based on the statistical process of Welch׳s periodogram of the stator currents in order to obtain stable and normalized fault indicators. The principle of the method is to statistically compare the current spectrum to a healthy reference so as to quantify the changes over the time. A statistic-based indicator is then constructed by monitoring specific harmonic family. The proposed method was tested on two experimental test campaigns for four different speeds and compared to a vibration indicator. The method was evaluated using a rigorous performance evaluation metric. A threshold evaluation was performed and shows that the proposed method is very tolerant to the machine speed. Thus, the use of a unique fault threshold whatever the speed can be considered. Results showed excellent agreement as compared with the vibration indicator, with an overall correlation of r=0.74 and only 4% of false alarms. Performance demonstrated by this novel method was superior to those of a classical energy-based indicator in terms of correlation with the vibration indicator and detection stability. Moreover, results also showed a better robustness of the proposed method since good performance can be obtained with the same detection threshold whatever the speed or the measure campaign whereas it needs to be redefined for each case with the classical indicator. This work shows the advantages of a statistic-based approach in order to increase the robustness of bearing fault detection in permanent-magnet synchronous machines. [Copyright &y& Elsevier]

Abstract: Due to limited information given by traditional local statistics, a new statistical modeling method for rolling element bearing fault signals is proposed based on alpha–stable distribution. In order to fully take advantages of complete information provided by alpha–stable distribution, this paper focuses on testing the validity of the proposed statistical model. A number of hypothetical test methods were applied to practical bearing fault vibration signals with different fault types and degrees. Through testing on the consistency of three alpha–stable parameter estimation methods, and the probability density function fitting level between fault signals and their corresponding hypothetical alpha–stable distributions, it can be concluded that such a non-Gaussian model is sufficient to thoroughly describe the statistical characteristics of bearing fault signals with impulsive behaviors, and consequently the alpha–stable hypothesis is verified. In the meantime, a new bearing fault detection method based on kurtogram and α parameter of the alpha–stable model is proposed, experimental results have shown that the proposed method has better performance on detecting incipient bearing faults than that based on the traditional kurtogram. [Copyright &y& Elsevier]

Abstract: Bearings are among the most frequently used components. Bearing failure could lead to complete stall of a mechanical system, unpredicted productivity loss for production facilities or catastrophic consequence for mission-critical equipment. As such, bearing fault detection and diagnosis is an imperative part of most of preventive maintenance procedures. This paper presents a parameter independent yet simple to implement fault detection technique. The Teager energy operator is tailored to extract both the amplitude and frequency modulations of the vibration signals measured from mechanical systems. The incorporation of the frequency modulation information into the proposed bearing fault detection method has eliminated the need for interference removal steps. As the amplitude demodulation (AD) is also inherent in the energy operator, the fault frequency can be detected from the spectrum of the energy-transformed signal. The effectiveness of the proposed method has been validated using both simulated and experimental data. [Copyright &y& Elsevier]