Bearing Fault Detection Research Articles

Efficient vibration analysis system using empirical mode decomposition residual signal and multi-axis data

The empirical mode decomposition (EMD) method is a technique that recursively decomposes an input signal into intrinsic mode functions (IMFs) by residual signals, primarily for identifying desirable features. The suggested algorithm observes the residual signal instead of the IMF, which lowers the computing load. The study introduces a new method for detecting bearing faults by enhancing signal extraction from sensor data using EMD and multi-axis feature extraction. This method streamlines the process by filtering out high-frequency noise and correlating residual signal information with analysis. The approach also enhances the signal-to-noise ratio (SNR) and feature signature identification using digital signal processing (DSP) techniques. The algorithm for vibration data analysis is tested for bearing failures, identifying shaft frequency and inner race bearing faults, which can be implemented in parallel. For the inner race fault bearing analysis, two-level EMD with a residual signal generates output similar to five-iteration EMD, saving 60% of computations. The use of spectral multiplication to multi-axis data processing produced a rise in the SNR of 18.32 dB to 20.92 dB for Y-axis and X-axis input, respectively. When compared to the single-axis IMF data computation, 20% fewer iterations are needed overall. A single-level EMD is adequate for calculating the rotational frequency of a healthy bearing. For the Y- and X-axis input, multi-axis analysis increases SNR by 10.68 dB and 13.14 dB, accordingly. This comprehensive strategy reduces computational complexity, improves fault detection accuracy, and minimizes noise impact, making it a promising solution for bearing fault detection.

A novel comprehensive method for weak signal blind detection based on adaptive quasi-periodic potential stochastic resonance and its application in bearing fault detection

In order to solve the common output saturation of stochastic resonance systems and the limitation of classical index SNR for blind detection, a novel adaptive quasi-periodic potential stochastic resonance blind detection method is proposed. First, a model of quasi-periodic potential stochastic resonance (QPPSR) possessing infinite steady state is constructed and analyzed for its structure change pattern. The superior performance of the model is verified by using the fourth-order Runge–Kutta algorithm. Secondly, the mechanism of QPPSR is analyzed using the probability flow method, which reveals the relationship between system parameters and performance. Again, a novel comprehensive blind detection index (CBDI) is exquisitely constructed to make up for the shortcomings of each indicator. Finally, CBDI and QPPSR are constructed into an adaptive blind detection system and applied to bearing fault detection. The results analyzed by experiments verify the good engineering application prospect of CBDI-QPPSR.

Fault diagnosis algorithm based on GADF-DFT and multi-kernel domain coordinated adaptive network

To address the problems of low detection accuracy of rolling bearings under different loads and the difficulty of effectively identifying the lack of labelled data, a rolling bearing fault diagnosis method combining GADF-DFT image coding and Multi-kernel domain coordinated adaptation network is proposed. Firstly, the vibration signal is converted into a two-dimensional image using GADF coding technology, and then the GADF image is converted into the frequency domain using discrete Fourier transform to extract deeper feature information. Combined with the multi-source domain adaptive method, the public feature extraction module is used to initially achieve feature mining of the image; the MK-MMD algorithm of the domain-specific adaptive module reduces the difference in feature distribution between the source and target domains; and the final classification difference minimization module reduces the problems caused by the classification errors that may be generated by the different domain classifiers due to the fact that the data samples are located near the category boundaries. The test uses the Case Western Reserve University dataset and divides the dataset with different operating conditions as the source and target domains, and the test results show that the proposed model demonstrates its effectiveness in responding to the complex operating condition changes in rolling bearing fault detection in multiple operating condition migration tasks, good adaptability and robustness, and is able to achieve accurate fault diagnosis under different operating conditions.

Application of Adaptive Lasso-Based Minimum Entropy Deconvolution for Bearing Fault Detection Based on Vibration Signal

Application of Adaptive Lasso-Based Minimum Entropy Deconvolution for Bearing Fault Detection Based on Vibration Signal

A shunted-swin transformer for surface defect detection in roller bearings

Although surface defect detection for roller bearings has received much attention in smart industrial manufacturing, most existing approaches employed the existing Convolution or Transformer models without modification to detect the defects. Such applications are undesirable because the classification models lack the necessary spatial relationship between multiple objects within an image. To develop a specialized model to detect bearing defects, we expand the applicability of the Swin Transformer. This paper presented a shunted-window Transformer (sSwin), which replaces the shifted-window mechanism with the shunted large–small windows. In sSwin, the self-attention heads can capture multi-scale token granularity features within the same self-attention layer. Also, a proposed local connection module enhances the boundary interaction of adjacent value tokens. As a result of the different-scale receptive fields of tokens within one attention layer, the model can bring significant performance gain on the defective detection tasks. The experiment demonstrates that sSwin-T outperforms the classical Swin-T by +2.8% APb and +1.9% APm, and Swinv2-T by +1.0% APb and +1.1% APm on COCO. It also surpasses them by +1.1% APb and +3.4% APb on the private bearing dataset with Cascade Mask-RCNN 3x.

Smart systems for real-time bearing faults diagnosis by using vibro-acoustics sensor fusion with Bayesian optimised 1-D CNNs

ABSTRACT Diagnosis of bearing faults in real-time is challenging when healthy bearing conditions are mixed with faulty ones, affecting the overall system of rotating machinery. Deep Learning provides an effective approach for condition-based maintenance of bearing faults, bypassing traditional signal processing complexity. Convolutional Neural Networks (CNNs) excel in real-time fault detection in bearings, leveraging their feature extraction capabilities from heterogeneous sensors. Usually, selecting optimal hyperparameters for CNNs is time-consuming and impacts model performance. Recent literature demonstrates that CNNs-based models for detecting bearing faults typically undergo trial searches to select optimal hyperparameters, leading to time-intensive procedures. To fill this research gap, our study proposes a Bayesian optimised 1-D CNNs method to address hyperparameter tuning challenges. Using Machinery Fault Simulator®, we demonstrate the effectiveness of our approach in identifying various bearing fault conditions through vibro-acoustics sensors. Bayesian optimisation efficiently partitions datasets for parallel computation, optimises hyperparameters, and minimises loss functions to enhance validation accuracy. The proposed method achieved a test accuracy of 99.62%, surpassing the benchmark’s 99.27%. Its effectiveness for bearing fault diagnosis is evident, compared to 96.76% without optimisation. Therefore, this study presents technical innovations, showcasing the diagnosis of diverse bearing faults with limited data through the integration of vibro-acoustics sensors.

Bearing race fault detection using an optomechanical micro-resonator

Bearing fault detection plays a crucial role in ensuring machinery reliability and safety. However, the existing bearing-fault-detection sensors are commonly too large to be embedded in narrow areas of bearings and too vulnerable to work in complex environment. Here, we demonstrate an approach to distinguish the presence of race faults in bearings and their types by using an optomechanical micro-resonator. The principle of the amplitude-frequency modulation model mixing fault frequency with mechanical frequency is raised to explain the asymmetrical sideband phenomena detected by the optical microtoroidal sensor. Kurtosis estimation used in this work can distinguish normal and faulty bearings in the time domain with the maximum accuracy rate of 91.72% exceeding the industry standard rate of 90%, while the amplitude-frequency modulation of the fault signal and mechanical mode is introduced to identify the types of the bearing faults, including, e.g., outer race fault and inner race fault. The fault-detection methods have been applied to the bearing on a mimic unmanned aerial vehicle (UAV), and correctly confirmed the presence of fault and the type of outer or inner race fault. Our study gives new perspectives for precise measurements on early fault warning of bearings, and may find applications in other fields such as vibration sensing.

Exploring a Cutting-Edge Framework for Bearing Fault Detection: A Synergistic Approach Integrating Statistical Analysis and Deep Learning Methods

Exploring a Cutting-Edge Framework for Bearing Fault Detection: A Synergistic Approach Integrating Statistical Analysis and Deep Learning Methods

Bearing-DETR: A Lightweight Deep Learning Model for Bearing Defect Detection Based on RT-DETR.

Detecting bearing defects accurately and efficiently is critical for industrial safety and efficiency. This paper introduces Bearing-DETR, a deep learning model optimised using the Real-Time Detection Transformer (RT-DETR) architecture. Enhanced with Dysample Dynamic Upsampling, Efficient Model Optimization (EMO) with Meta-Mobile Blocks (MMB), and Deformable Large Kernel Attention (D-LKA), Bearing-DETR offers significant improvements in defect detection while maintaining a lightweight framework suitable for low-resource devices. Validated on a dataset from a chemical plant, Bearing-DETR outperformed the standard RT-DETR, achieving a mean average precision (mAP) of 94.3% at IoU = 0.5 and 57.5% at IoU = 0.5-0.95. It also reduced floating-point operations (FLOPs) to 8.2 G and parameters to 3.2 M, underscoring its enhanced efficiency and reduced computational demands. These results demonstrate the potential of Bearing-DETR to transform maintenance strategies and quality control across manufacturing environments, emphasising adaptability and impact on sustainability and operational costs.

Iterative Laplace of Gaussian filter and improved Teager energy operator for bearing fault detection in gearboxes

Abstract The presence of cyclic impulses is considered to be a significant indicator of bearing defects. The extraction of fault-related information from gearboxes, however, poses a significant challenge due to the complex working conditions in gear transmission systems. These conditions often lead to fault-related impulse features being masked by inevitable background noise and vibration components. To address this issue, a novel iterative Laplace of Gaussian (LoG) filtering technique is proposed for the fault diagnosis field of bearings in gearboxes. An objective function incorporating Kurtosis statistics is used to iteratively update LoG coefficients instead of relying on fixed coefficients as in the original version, making the iterative LoG filtering technique better suited for handling fault-induced impact signals. Furthermore, an improved Teager energy operator (TEO) is developed to reduce the susceptibility of the original TEO to noise. The improvement involves incorporating a multi-resolution symmetric difference sequence into the discrete expression of TEO. This symmetric difference acts as a translation-invariant, sliding-window filter that, when combined with an energy operator, enables both energy detection and frequency filtering. The robustness against noise and sensitivity to frequency can be further improved by adjusting the multi-resolution parameter. The results of the experiments confirm the effectiveness of the proposed approach in identifying weak bearing fault features for gearboxes. Besides, the proposed method is also compared with the original LoG filter-based method and some competing methods, demonstrating its superior performance.

Cyclostationarity and real order derivatives in roller bearing fault detection

Abstract Various methods are used in the field of machine diagnostics for recognizing cyclostationarity in signals. The real order derivatives of vibration signals, however, have been rarely reported from the perspective of their effect on the performance of cyclostationarity detection methods. In this paper, we use real order derivatives together with spectral correlation, spectral coherence and squared envelope. Our results suggest that adjusting the order of derivative can enhance the analysis outcome of spectral correlation and squared envelope in particular. Remarkably, the results also suggest that squared envelope, when used alongside real-order derivatives, may replace spectral correlation and spectral coherence. This approach allows obtaining results with reduced computational power, making it advantageous for applications like industrial edge computing, where cost-effective hardware is crucial.

Research and Application of Two-Dimensional Time-Delayed Tri-Stable Stochastic Resonance System for Bearing Fault Detection

The time-delayed feedback term can improve the output of a system, while the two-dimensional stochastic resonance (SR) system has a stronger signal amplification capability. To improve the output signal-to-noise ratio (SNR) of the system, this paper proposes a two-dimensional time-delayed tri-stable stochastic resonance system (TDTDTSR) based on the advantages of the above two systems. First, the steady-state probability density function (SPD), the mean first-pass time (MFPT), and the output SNR are derived under adiabatic approximation theory, and the effects of different system parameters on them are investigated. Next, TDTDTSR and the classical two-dimensional tri-stable stochastic resonance system (CTDTSR) system are simulated numerically. The results show that the mean signal-to-noise gain (MSNRG) of TDTDTSR system is higher than that of the CTDTSR system. Finally, the system parameters are optimized using a genetic algorithm, and the application of TDTDTSR to bearing fault detection is compared with CTDTSR and the novel piecewise symmetric two-dimensional tri-stable stochastic resonance (NPSTDTSR) systems. The experimental results demonstrate that TDTDTSR system has better performance, providing valuable theoretical support and practical engineering applications for the system in subsequent analyses.

Methodology for the Detection of Contamination and Gradual Outer Race Faults in Bearings by Fusion of Statistical Vibration–Current Features and SVM Classifier

Bearings are one of the main components of induction motors, machines widely employed in today’s industries, making their monitoring a primordial task; however, most systems focus on measuring one physical magnitude to detect one kind of fault at a time. This research tackles the combination of two common faults, grease contamination and outer race damage, as lubricant contamination significantly impacts the life of the bearing and the emergence of other defects; as a contribution, this paper proposes a methodology for the diagnosis of this combination of faults based on a proprietary data acquisition system measuring vibration and current signals, from which time domain statistical and fractal features are computed and then fused using LDA for dimensionality reduction, ending with an SVM model for classification, achieving 97.1% accuracy, correctly diagnosing the combination of the contamination with different severities of the outer race damage, improving the classification results achieved when using vibration and current signals individually by 7.8% and 27.2%, respectively.

Delay Segmented Tristable Stochastic Resonance System Driven by Non-Gaussian Colored Noise and Its Application in Bearing Fault Detection

Abstract This study investigates the Delayed Segmented Tristable Stochastic Resonance (DSTSR) system under the influence of additive non-Gaussian colored noise. The research employs an improved segmented tristable potential function, wherein the equilibrium points and barrier heights can be independently controlled by parameters. Simultaneously, the segmented function on both sides reduces the restrictions of higher-order terms on the walls of the potential wells. The equivalent Langevin equation for the DSTSR system is obtained using the path integral method, the unified colored noise approximation method, and the small-delay approximation. Subsequently, the theoretical expressions for the steady-state probability density, mean first passage time (MFPT), and Signal-to-Noise Ratio (SNR) are derived from the resulting equations, and the impact of variations in system parameters on these performance metrics is discussed. Additionally, Monte Carlo simulations for MFPT are conducted to verify the accuracy of the theoretical derivations. Combining the results from the theoretical section and the impact of parameters on system performance, the article employs an adaptive genetic algorithm to optimize system parameters. This algorithm is then applied to simulation experiments and bearing fault detection. In the simulation experiments, the DSTSR system is compared with other systems. The results indicate that the DSTSR system exhibits the highest SNR improvement. Furthermore, in bearing fault detection under non-Gaussian colored noise, the DSTSR system shows higher spectral amplitude and SNR at the fault frequency compared to the tristable stochastic resonance system and the segmented tristable stochastic resonance system without time delay feedback. This suggests that stochastic resonance can effectively detect weak signals in non-Gaussian non-white noise scenarios, and the introduction of time delay contributes to the occurrence of stochastic resonance to a certain extent.

CAC-YOLOv8: real-time bearing defect detection based on channel attenuation and expanded receptive field strategy

Bearing defect detection plays a crucial role in the intelligent production of chemical transmission equipment, where timely identification and handling of defective bearings are essential. However, in practical large-scale industrial production, product surface defects are often complex, diverse, and exhibit significant variations in appearance, posing severe challenges to the discriminative ability and detection efficiency of bearing defect detection algorithms. This paper proposes a real-time bearing surface defect detection algorithm, CAC-YOLOv8, which designs the Channel Attenuation Network (CAN) and Compound Pooling Pyramid Spatial Pyramid Pooling Fast (CPPSPPF) structure. Specifically, the model introduces the Channel Attenuation Network to achieve parallel feature extraction, deep feature processing, and feature fusion under different channel numbers, capturing critical features related to bearing defects and thereby improving the inference speed. Subsequently, based on the concept of overlapped receptive fields, a CPPSPPF structure is constructed, utilizing multiple iterations of max-pooling operations with smaller pooling kernel sizes to prevent information loss while expanding the receptive field, thereby strengthening the capturing ability of features at different scales. The experimental results indicate that the proposed CAC-YOLOv8 bearing surface defect detection algorithm, compared to the YOLOv8 model, achieved a 0.3% improvement in mAP@0.5, reduced model size by 14.4%, and enhanced model inference speed by 33.3%. This enables the CAC-YOLOv8 model to significantly improve the real-time performance of bearing defect detection while maintaining high-precision detection. The performance in practical industrial detection demonstrates that the proposed approach has achieved outstanding results in both speed and accuracy.

Cross-condition bearing fault detection based on online drift detection and domain adaptation

Aiming at the problem that the data distribution of bearings across operating conditions generates offset resulting in insufficient diagnostic accuracy of the original model for new data, a cross-condition bearing fault detection method based on online drift detection and domain adaptation is proposed. First, the original one-dimensional vibration signals collected are transformed by a two-dimensional wavelet transform to convert the time-frequency image dataset. Second, the drift detection of the data across operating conditions is carried out using Random Forest (RF), and the 3σ criterion as well as the drift detection judgment criteria are set. Next, the source domain model based on Googlenet is used to extract features from the target domain data, and the Whale Optimization Algorithm to Improve Local Preserving Projection Algorithm (WOA-LPP) algorithm is combined to construct a brand-new projection space to align the features of the source and target domains. Then, the source and target domain features are reconstructed by combining the LPP optimal projection matrix to construct a fully connected network trained by the source domain features. Finally, probabilistic label-based decision fusion is proposed to integrate multiple classifiers to reduce the effects of model training randomness and strong noise interference. Validated by the publicly available Western Reserve University bearing data, the method proposed in this paper has good detection accuracy as well as robustness across operating conditions, which can effectively improve the defects of shifting data distribution and degradation of model accuracy under variable speed.

A noise-robust CNN architecture with global attention and gated convolutional Kernels for bearing fault detection

Detecting faults in bearings is essential for the maintenance and operation of rotating machinery. However, achieving high accuracy and noise immunity is challenging due to the involvement of intricate and noisy signals. To address this issue, this paper introduces a multi-scale separable gated convolutional neural network (GCK-MSSC). In the GCK-MSSC model, the gate convolutional kernel replaces the conventional convolutional kernel. It is designed to dynamically adjust the convolution kernel’s weights based on the input features. Additionally, the one-dimensional global attention mechanism is incorporated, enhancing the model’s global awareness within the MSSC framework. The experimental results on two public bearing datasets confirm the performance of the proposed method. It demonstrates improved performance over current leading-edge methods, especially in terms of accuracy, and proves to be significantly robust against various levels of noise. Specifically, it achieves accuracies of 99.45% and 99.78% on the two datasets. Furthermore, even after the addition of noise with a signal-to-noise ratio of 0, it still maintains an accuracy as high as 85.65% (on the Politecnico di Torino dataset).

Bearing fault detection using a method involving absolute value spectrum and impulsivity evaluation

Bearing fault detection using a method involving absolute value spectrum and impulsivity evaluation

Early-Stage Fault Diagnosis of Motor Bearing Based on Kurtosis Weighting and Fusion of Current-Vibration Signals.

To solve the problem of a low signal-to-noise ratio of fault signals and the difficulty in effectively and accurately identifying the fault state in the early stage of motor bearing fault occurrence, this paper proposes an early fault diagnosis method for bearings based on the Differential Local Mean Decomposition (DLMD) and fusion of current-vibration signals. This method uses DLMD to decompose the current signal and vibration signal, respectively, and weights the decomposed product function (PF) according to the kurtosis value to reconstruct the signal, and then fuses the reconstructed signals to obtain the current-vibration fusion signal after normalization, and then analyzes the fusion signal spectrally through the Hilbert envelope spectrum. Finally, the fusion signal is analyzed by the Hilbert envelope spectrum, and a clear fault characteristic frequency is obtained. The experimental results demonstrate that compared to traditional bearing fault diagnosis methods, the proposed method significantly improves the signal-to-noise ratio of fault signals, effectively enhances the sensitivity of early-stage fault detection in motor bearings, and improves the accuracy of fault identification.

Enhanced Bearing Fault Diagnosis Through Trees Ensemble Method and Feature Importance Analysis

Abstract Purpose This research introduces a groundbreaking method for bearing defect detection. It leverages ensemble machine learning (ML) models and conducts comprehensive feature importance analysis. The key innovation is the training and benchmarking of three tree ensemble models—Decision Tree (DT), Random Forest (RF), and Extreme Gradient Boosting (XGBoost)—on an extensive experimental dataset (QU-DMBF) collected from bearing tests with seeded defects of varying sizes on the inner and outer raceways under different operating conditions. Method The dataset was meticulously prepared with categorical variable encoding and Min–Max data normalization to ensure consistent class distribution and model accuracy. Implementing the ML models involved a grid search method for hyperparameter tuning, focusing on reporting the models’ accuracy. The study also explores applying ensemble methods and using supervised and unsupervised learning algorithms for bearing fault detection. It underscores the value of feature importance analysis in understanding the contributions of specific inputs to the model’s performance. The research compares the ML models to traditional methods and discusses their potential for advanced fault diagnosis in bearing systems. Results and Conclusions The XGBoost model, trained on data from actual bearing tests, outperformed the others, achieving 92% accuracy in detecting bearing health and fault location. However, a deeper analysis of feature importance reveals that the models weigh certain experimental conditions differently—such as sensor location and motor speed. This research’s primary novelties and contributions are comparative evaluation, experimental validation, accuracy benchmarking, and interpretable feature importance analysis. This comprehensive methodology advances the bearing health monitoring field and has significant practical implications for condition-based maintenance, potentially leading to substantial cost savings and improved operational efficiency.