Search

Deep-learning-based identification and diagnosis of faults in industrial assets has received much attention. It does not need a deep understanding of the target domain or complex signal-processing techniques for time-consuming feature identification and selection. The vast majority of approaches developed in this field rely on public data recorded in laboratory conditions with high-performance measurement equipment. Moreover, primarily vibration data is used for condition monitoring tasks, which is particularly sensitive to typical error patterns in rotating machinery. These conditions are difficult to maintain in industrial environments since using high-performance measurement systems would not be economically feasible. This paper demonstrates that modern deep learning can achieve extraordinary fault detection results with industrial standard low-cost measurement systems. We determine measurements taken from the industrial-sensory equipment that potentially capture the system's fault patterns. Hereafter, we apply traditional failure detection and identification techniques to evaluate the suitability of the taken measurements for bearing fault detection. These techniques apply envelope analysis to reveal the fault patterns in the signals and a support vector machine to separate the fault classes. We can conclude that the used low-cost sensory equipment is capable to capture meaningful, fault-describing patterns. To improve the failure detection performance we are investigating a 2D, as well as 1D, convolutional neural network approach to identify the error patterns and classify the respective errors. We compare the deep-learning-based methods with the traditional methods. Furthermore, we assess which inverter signal carries the largest fault-describing information content. Experimental results indicate that the proposed deep learning methods outperform traditional fault diagnosis methods, hence, demonstrating the effectiveness in an industrial condition monitoring application. [ABSTRACT FROM AUTHOR]

Abstract Transient impulse contains abundant information of bearings status. When fault occurs, it is activated and would recur periodically or quasi-periodically. Its period can indicate where defects lie in. However, transient impulse is easily swallowed by background noise or interferences in part or in whole, especially at early stage of fault. This problem brings hard obstacles into faults detection. Considering that transient impulses are periodical or quasi-periodical and vibration signal has local similarity, the single transient impulse can be seen as one of shift-invariant features. In view of this, this paper derives adaptive and non-linear signal decomposition formulas and further proposes adaptive and unsupervised feature learning method by using convolutional restricted Boltzmann machine model. With respecting local waveform structures, this method can automatically capture shift-invariant patterns hidden in original signal and decompose the original signal into several sub-components at the cost of minimizing reconstruction error. Among these sub-components, the fault-related information, i.e., transient impulses signal, could be extracted likely. It provides a promising idea for intelligent signal processing by using unsupervised learning. Afterwards, Maximizing kurtosis is applied to select optimally latent fault component. Two real bearing experiments validate this method is effective and reliable in extraction of weak transient impulses. [ABSTRACT FROM AUTHOR]

Vibration signal analysis is a significant approach in fault diagnosis. This article presents a new vibration signal preprocessing scheme called Automated Relative Energy based Empirical Mode Decomposition (AREEMD) and a novel Cepstral Autoregressive (cAR) feature extraction. AREEMD generates the modes based on high relative energies and automates the signal reconstruction. The cAR features take advantage of the cepstrum information and provide discriminant markers for various machine faults. The weighted K-nearest neighbor delivered the best results when validated on four different datasets. The proposed scheme yielded promising performances of average accuracies of 99.3% using SUBF, 97.4% for CWRU, 100% on the Ottawa, and 99.1% using the combined dataset. The significant generalizability of the technique is demonstrated by its high performance in identifying faults that will aid in the detection of early warnings, regardless of motor speeds and size of faults, assuring a reduction in system shutdowns brought on by bearing failure. [Display omitted] • A new in-house machine fault vibration dataset (SUBF v1.0) is created. • A novel AREEMD technique is devised for intelligent vibration signal preprocessing. • Novel powerful discriminative Cepstral Autoregressive Features are proposed. • Proposed framework is validated on SUBF, CWRU, Ottawa, and a combined dataset. [ABSTRACT FROM AUTHOR]

This paper presents a new and straightforward system for bearing fault detection. The system computes the stability of two vibration signals by using the direct matching points (DMP) of an elastic and non-linear align function. It is able to find discriminant properties in the stability of fault-free and faulty bearing vibration signals from the early and late stages of the fault in critical bearing parts. Because training data constitutes one of the critical challenges in most expert and intelligent systems, one of the novelties of the proposed stability-based system is that it requires neither training nor fine-tuning. A significant impact on the robustness of the system is demonstrated using two publicly available vibration signal databases under several load conditions, with real faults, during multiple machine working states. Experimental results validate the use of the proposed stability-based system for predictive maintenance in bearings. [ABSTRACT FROM AUTHOR]

Condition monitoring of electric drives is of paramount importance since it contributes to enhance the system reliability and availability. Moreover, the knowledge about the fault mode behavior is extremely important in order to improve system protection and fault-tolerant control. Fault detection and diagnosis in squirrel cage induction machines based on motor current signature analysis (MCSA) has been widely investigated. Several high resolution spectral estimation techniques have been developed and used to detect induction machine abnormal operating conditions. This paper focuses on the application of MCSA for the detection of abnormal mechanical conditions that may lead to induction machines failure. In fact, this paper is devoted to the detection of single-point defects in bearings based on parametric spectral estimation. A multi-dimensional MUSIC (MD MUSIC) algorithm has been developed for bearing faults detection based on bearing faults characteristic frequencies. This method has been used to estimate the fundamental frequency and the fault related frequency. Then, an amplitude estimator of the fault characteristic frequencies has been proposed and fault indicator has been derived for fault severity measurement. The proposed bearing faults detection approach is assessed using simulated stator currents data, issued from a coupled electromagnetic circuits approach for air-gap eccentricity emulating bearing faults. Then, experimental data are used for validation purposes. [ABSTRACT FROM AUTHOR]

• A graph similarity guided symplectic geometry mode decomposition (GS-SGMD) based dynamic time warping to detect bearings faults. • GS-SGMD is applied to extract fault features from raw vibration signals. • Dynamic time warping is used to realize simple and fast classification. • The performance of the method is validated by experimental investigations to detect faults in bearings. A bearing's health state is closely linked to the reliable operation of rotating machinery. In this context, dynamic time warping (DTW) is an excellent fault classifier due to its outstanding distance measurement ability. However, DTW alone cannot obtain acceptable results when it is employed to handle signals with a certain degree of noise. An enhancement of DTW based on graph similarity guided symplectic geometry mode decomposition (GS-SGMD) is presented in this paper to improve the performance of traditional DTW. To reduce the interference of random noise in raw signals, the signal is decomposed into multiple components by SGMD. Then, graph similarity is introduced to select the effective component as a test sample to be detected. In addition, the templates of known fault states of bearings are also obtained by GS-SGMD. Finally, DTW is employed to recognize the fault type of the test sample. Experimental results show that the presented method can effectively detect bearing faults with higher precisions in comparison to DTW and SGMD. Flow-chart of the presented method (GS-SGMD-based DTW). [Display omitted] [ABSTRACT FROM AUTHOR]

In this paper, we propose an intelligent bearing fault detection method based on a calculus enhanced energy operator (CEEO). The main purpose is to extract the bearing fault signatures in the presence of strong noise and multiple vibration interferences without prior information of the resonance excited by the bearing fault. This new energy operator exploits both the interference handling capability of a differentiation step and the noise suppression nature of the integration process. It also shares the simplicity, computational efficiency, and the ability to reveal signal impulsiveness of the energy operator. All these elements, i.e., differentiation, integration and energy operator, are implemented by a simple formula in a single step. Another advantage of the CEEO method is that, unlike the popular high frequency resonance methods, it does not require a bandpass filtering step and hence eliminates the burden to acquire the resonance information. As such, it is suited to on-line bearing fault detection in a noisy environment with multiple vibration interferences. Our simulation studies have shown that the CEEO method outperforms the conventional energy operator and the enveloping methods in handling both noise and interferences. Its performance has also been examined using our experimental data and the data from the literature. [ABSTRACT FROM AUTHOR]

• A B-spline approximation algorithm, which can be considered a noise reduction method, is proposed in this paper. • An alternative energy operator named envelope-derivative operator and the designed B-spline approximation are integrated into a new energy measure for extracting the bearing fault signatures. • It is evaluated using numerical simulated and experimental results. The results demonstrate that the proposed method is robust and efficient. The vibration fault signal recorded from a faulty bearing is generally non-stationary in nature and the envelope of such a signal is usually time-varying. This may bring challenges to bearing fault detection. To overcome such difficulties, a novel energy demodulation technique is proposed in this work, which first interpolates the raw signals with nonuniform B-spline approximation and then utilizes an energy demodulation tool based on an enhanced energy measure called envelope-derivative operator (EDO). The B-spline approximation and the EDO are integrated into a new energy operator for extracting the fault signatures. The B-spline EDO enjoys high time resolution and performs well in the case of strong noise. The superiority of the spline-based EDO in dealing with strong noise is demonstrated by simulation and real experiments. [ABSTRACT FROM AUTHOR]

Abstract: Traditional envelope analysis must examine all the resonant frequency bands during the process of bearing fault detection. To eliminate the above deficiency, this paper presents an insight concept based on the empirical mode decomposition to choose an appropriate resonant frequency band for characterizing feature frequencies of bearing faults by using the envelope analysis subsequently. By the band-pass filtering nature of the empirical mode decomposition, the resonant frequency bands are allocated in a specific intrinsic mode function. The inner or outer ring of bearings scratched intentionally is used to validate the feasibility of the proposed idea, and comparisons with the traditional envelope analysis are addressed. The experimental results show that the proposed insight concept can efficiently and correctly diagnose the bearing fault types. [Copyright &y& Elsevier]

• Proposed cycle envelope spectral segmentation method based to optimize EWT. • Calculate the similarity of the envelope spectra of adjacent signals to optimize the components. • A new indicator called sensitive IMFs assessing index (SIAI) was designed to identify fault information. The empirical wavelet transform (EWT) has made outstanding contributions in the field of fault diagnosis. However, when EWT processes complex signals, there may be modal aliasing or meaningless components. In order to solve the spectral segmentation defect of the EWT method and improve its ability to extract bearing fault features, this paper proposes a new algorithm to improve EWT. In order to weaken the influence of extreme points in the complex Fourier spectrum on modal differentiation, the cycle envelope spectral segmentation method is proposed. The maximum envelope fitting method not only reduces the number of useless extreme points, but also highlights each mode. The number of filters is reduced while suppressing the interference of noise on the modal. The components that contain similar information in the result will be merged based on their correlation. Then sensitive IMFs assessing index (SIAI) is proposed to combine correlation and symmetry of the signal, and obtains the characteristics of time domain, frequency domain, and amplitude domain to extract fault information. Locomotive bearing and wind turbine gearbox bearing fault data verified the effectiveness of the method. [ABSTRACT FROM AUTHOR]

Abstract: Most of the difficulties encountered when detecting initial bearing defects are due to the presence of noise in the measured signals. In the present paper, we show the effectiveness of using Adaptive Noise Cancellation techniques, in enhancing the vibration signal which is corrupted with noise in the low frequency spectrum. Two algorithms have been used in the paper, the Adaptive Noise Cancellation (ANC) and the Adaptive Self-Tuning filter (AST). The two algorithms have been first of all tested on a computer simulated signals; then, applied on measured signals on a bearing fault simulator. It has been found that adaptive filtering increases the signal to noise ratio for both simulated and measured signals. [Copyright &y& Elsevier]

Enhancing repetitive transients is a key to bearing fault detection. Blind deconvolution aims to recover impulsive impacts from repetitive transients and their mixture. Nowadays, based on different optimization criteria, such as kurtosis, entropy, generalized Lp / Lq norm, etc., various deconvolution methods have been proposed for recovering repetitive impacts. Due to the non-convex of these optimization criteria, local optimal solutions to these optimization problems may be obtained frequently. In this paper, inspired by the fast realization of spectral kurtosis, e.g. the fast kurtogram, an optimized minimum generalized Lp / Lq deconvolution (OMGD) is proposed and investigated. The main idea of the proposed method utilizes the filters designed in the fast kurtogram to provide proper initializations for the minimum generalized Lp / Lq deconvolution (MGD). Hence, multiple local optimal solutions can be iteratively obtained. Then, the minimum of these local optimal solutions is used as a precise estimate to approximate the globally optimal solution of MGD. Results show that the proposed method has better deconvolution performance than MGD and fast kurtogram for enhancement of repetitive impacts caused by localized bearing faults. [ABSTRACT FROM AUTHOR]

• Based on graph similarity evaluation, a two-stage method is proposed for bearing fault detection. • Graph similarity is verified as an effective utensil for the bearing operation monitoring. • With only few sampling points (100 samples in this research), the operation states can be rapidly identified. Robust identification of bearing health states is closely linked to timely condition monitoring and downtime reducing for rotating machinery. Although many proposed algorithms achieve extraordinary performances on feature extraction, uncertainty still remains for the bearing fault identification. To address this problem, this paper introduces a two-stage framework for bearing fault detection using graph similarity evaluation. The recognition stage is used to identify the operation state (fault or not) based on an improved graph-based method according to the sampled vibration signal for each spindle turn. The feature extraction stage, on the other hand, is implemented to extract the fault characters from the time-domain signals. The results indicate that the proposed method achieves 100% identification accuracy in bearing fault detection even with phase shifts. This work therefore provides a powerful tool for bearing faults detection and is broadly applicable to a variety of engineering applications and experimental conditions. [ABSTRACT FROM AUTHOR]