Information For Fault Diagnosis Research Articles

MSTKernel Net: a rolling bearing intelligent diagnosis framework based on short-time time–frequency convolution

Abstract Monitoring rotating machinery is a key task in modern production processes. The emergence of deep learning technology has significantly improved the performance of intelligent diagnosis systems for such machinery. However, despite the commendable performance of many existing frameworks, they lack transparency, which hinders their interpretability in fault diagnosis based on directional signals. This study addresses this challenge by delving into the fault features present in vibration signals and designing a convolutional module specifically tailored to these characteristics, modularized short time–frequency kernel (MSTKernel). This innovative framework, MSTKernel Network, employs convolutional neural networks for feature extraction, simulating the time–frequency sliding process through convolutional properties while preserving temporal features and enriching fault diagnosis information. Through experimental data testing and visualization of convolutional kernel characteristics, we evaluate the potential of this framework to significantly enhance the fault diagnosis capability of rolling bearings, demonstrating its practicality and effectiveness in real-world applications.

Gaussian bistable cascade double feedback stochastic resonance weak signal enhancement detection

Stochastic resonance (SR), as a nonlinear weak signal enhancement detection method, has found extensive applications in fields like fault diagnosis and biological information processing. However, further research has revealed that single stochastic resonance systems often fail to meet the requirements of practical engineering applications, leading to a growing interest in multi-system collaborative stochastic resonance. This article first constructs a cascaded SR system based on the Gaussian bistable model, and introduces the idea of feedback into the cascaded system. A Gaussian bistable cascaded dual feedback SR system is proposed to improve the output signal enhancement performance of the system. This system effectively utilizes the memory characteristics inherent in the output feedback and enables multi-level particle transitions and energy transfers in noisy input signals, thereby reducing noise and enhancing the extraction of weak signal features. Furthermore, the BSO optimization algorithm was utilized to optimize the system parameters of the model, resulting in an adaptive Gaussian bistable cascaded double feedback stochastic resonance (GBCDFSR) system. This system is applied to the processing of noisy periodic signals across various noise types. Simulation experiments and engineering applications show that the proposed model has stronger signal detection capabilities.

Gearbox Fault Diagnosis Based on MSCNN-LSTM-CBAM-SE.

Ensuring the safety of mechanical equipment, gearbox fault diagnosis is crucial for the stable operation of the whole system. However, existing diagnostic methods still have limitations, such as the analysis of single-scale features and insufficient recognition of global temporal dependencies. To address these issues, this article proposes a new method for gearbox fault diagnosis based on MSCNN-LSTM-CBAM-SE. The output of the CBAM-SE module is deeply integrated with the multi-scale features from MSCNN and the temporal features from LSTM, constructing a comprehensive feature representation that provides richer and more precise information for fault diagnosis. The effectiveness of this method has been validated with two sets of gearbox datasets and through ablation studies on this model. Experimental results show that the proposed model achieves excellent performance in terms of accuracy and F1 score, among other metrics. Finally, a comparison with other relevant fault diagnosis methods further verifies the advantages of the proposed model. This research offers a new solution for accurate fault diagnosis of gearboxes.

A periodic-modulation-oriented noise resistant correlation method for industrial fault diagnostics of rotating machinery under the circumstances of limited system signal availability

The periodical impulses caused by localized defects of components are the vital characteristic information for fault detection and diagnosis of rotating machines. In recent years, multitudinous spectrum analysis-based signal processing methods have been developed and authenticated as the powerful tools for excavating fault-related repetitive transients from the measured complex signals. Nonetheless, in practice, their applications can be severely confined by the constraints of limited system signal availability and incomplete information extraction under intricate noise interferences. To tackle the aforementioned issues, this paper proposes a periodic-modulation-oriented noise resistant correlation (PMONRC) method for target period detection and fault diagnosis of rotating machinery. Firstly, the envelope of raw signal is obtained via a novel sequential procedure of signal element-wise squaring, spectral Gini index-guided adaptive low-pass filtering, and signal element-wise square root computation, to highlight the modulated wave component that is more likely to be related to the potential fault-induced periods. Subsequently, a series of sub-signals, which can encode the fault-related repetitive information and enhance noise resistance, are constructed utilizing the envelope signal. Based upon the envelope signal and the obtained sub-signals, a weighted envelope noise resistant correlation function can be derived with the assistance of the L-moment ratio-based indicator and Sigmoid transformation. Finally, the specific fault type of the rotating machinery can be identified and affirmed accordingly. The proposed PMONRC method, which is nonparametric and completely adaptive to the signal being processed itself, overcomes the deficiencies of spectral analysis-based approaches, and is applicable for the engineering circumstances of system signal limitation and low signal-to-noise ratio (SNR), possessing immense practical merit. Both simulation analyses and experimental validations profoundly demonstrate that the proposed method is superior to other existing state-of-the-art time-domain correlation methods. Moreover, as an attempt as well as exemplar to apply this method, the PMONRC-based incipient fault diagnostic results of rolling bearing data from the well-known experimental platform PRONOSTIA are presented and discussed as well, to further elucidate the effectiveness and practical engineering significance of the proposed method.

Source-free domain adaptation method for fault diagnosis of rotation machinery under partial information

Fault diagnosis is crucial for reliability assessment of rotation machinery. Due to issues such as data privacy, it is impossible to get complete information for fault diagnosis in practical and challenging scenario. To solve aforementioned problem, fault diagnosis under partial information is studied. A source-free domain adaptation method for fault diagnosis, enabling cross-domain fault diagnosis without accessing the source data, is proposed. Firstly, multireceptive field graph convolutional(MRF-GCN) was used to aggregate different numbers of node information from different receptive fields for extracting more representative features. Secondly during the training process on the target domain, positive and negative pairs are constructed based on the samples’ neighbors and extended neighbors. Clustering and domain adaptation are then performed using a contrastive loss. Finally, information maximization loss is employed to improve the diagnostic accuracy. Experimental results demonstrate that, the proposed approach achieves favorable diagnostic performance under partial information, even without access to source domain data.

Image Enhancement Based on Dual-Branch Generative Adversarial Network Combining Spatial and Frequency Domain Information for Imbalanced Fault Diagnosis of Rolling Bearing

To address the problems of existing 2D image-based imbalanced fault diagnosis methods for rolling bearings, which generate images with inadequate texture details and color degradation, this paper proposes a novel image enhancement model based on a dual-branch generative adversarial network (GAN) combining spatial and frequency domain information for an imbalanced fault diagnosis of rolling bearing. Firstly, the original vibration signals are converted into 2D time–frequency (TF) images by a continuous wavelet transform, and a dual-branch GAN model with a symmetric structure is constructed. One branch utilizes an auxiliary classification GAN (ACGAN) to process the spatial information of the TF images, while the other employs a GAN with a frequency generator and a frequency discriminator to handle the frequency information of the input images after a fast Fourier transform. Then, a shuffle attention (SA) module based on an attention mechanism is integrated into the proposed model to improve the network’s expression ability and reduce the computational burden. Simultaneously, mean square error (MSE) is integrated into the loss functions of both generators to enhance the consistency of frequency information for the generated images. Additionally, a Wasserstein distance and gradient penalty are also incorporated into the losses of the two discriminators to prevent gradient vanishing and mode collapse. Under the supervision of the frequency WGAN-GP branch, an ACWGAN-GP can generate high-quality fault samples to balance the dataset. Finally, the balanced dataset is utilized to train the auxiliary classifier to achieve fault diagnosis. The effectiveness of the proposed method is validated by two rolling bearing datasets. When the imbalanced ratios of the four datasets are 0.5, 0.2, 0.1, and 0.05, respectively, their average classification accuracy reaches 99.35% on the CWRU bearing dataset. Meanwhile, the average classification accuracy reaches 96.62% on the MFS bearing dataset.

Adaptive time-step unscented kalman filtering (ATS-UKF) based observer design for urea selective catalytic reduction (SCR) performance of diesel engines

To reduce the number of sensors in the SCR catalyst, state feedback and fault diagnosis information are provided. Firstly, a model based on the coupling of flow, heat transfer, and gas-solid phase catalytic reaction in the SCR system is investigated in this paper. The parabolic partial differential equations are simplified by the variable substitution method and the method of lines approach (MOL). The simplified system of equations is solved by backward differentiation formulas (BDF) with adaptive adjustment time step strategy. Meanwhile, the chemical reaction parameters are accurately calibrated per second using the Levenberg-Marquardt method. Secondly, the ATS-UKF is designed in this paper, and to ensure the synchronisation between the ATS-UKF and the SCR model calculations, the time step of solving the BDF by the SCR model is taken as the time step of propagating the sigma points. Two observation scenarios are assumed: (1) no downstream NH3 concentration sensor, ammonia coverage and downstream NH3 concentration are observed by ATS-UKF; (2) no downstream NOx sensor, ammonia coverage and downstream NOx concentration are observed by ATS-UKF. Finally, the paper carries out bench tests. In the first case, the ammonia coverage obtained by the ATS-UKF reached 0.99 with respect to the model-calculated value R². The mean absolute error (MAE) between the observed and experimental values of the ATS-UKF for the downstream NH3 concentration was 2.76 ppm. In the second case, the ammonia coverage obtained by the ATS-UKF reached 0.99 with respect to the model-calculated value R², and the MAE between the observed and experimental values of the ATS-UKF for the downstream NOx concentration was 1.53 ppm. Environmental ImplicationThe Adaptive Time-Step Unscented Kalman Filtering (ATS-UKF) enhances urea Selective Catalytic Reduction (SCR) in diesel engines, improving environmental outcomes. This method minimizes sensor dependence, enabling more precise SCR system management and effective emission reduction. By advancing emission control technologies, ATS-UKF contributes to global air pollution mitigation efforts, supporting cleaner air and environmental sustainability. Its innovative approach in monitoring and predicting SCR performance marks a significant step towards eco-friendly diesel engine operation.

Fault prediction of electronic devices based on attention mechanism time-series point process

In recent years, with the development of high-precision sensors, computer technology and artificial intelligence, it makes the data-driven intelligent fault warning and diagnosis method gradually become a research hotspot. Aiming at the low utilization of fault diagnosis information of complex electronic devices and the problems of convergence and training time of traditional neural network fault diagnosis models, a point-in-time process generation model without using intensity function is proposed. The model uses the Wasserstein distance to construct the loss function, which facilitates the measurement of the deviation between the model distribution and the true distribution, and uses a self-concern mechanism to describe the degree of influence of the historical events on the current events, which makes the model interpretable and more capable of generalization. Comparative experiments show that without a priori information about the intensity function, the method reduces the relative error rate by 3.59 %, improves the fault prediction accuracy by 3.91 %, and has a better overall fit than the RNN-like generative model and the great likelihood model. Example analysis shows that the model has good prediction accuracy and provides a feasible solution for real-time fault diagnosis of complex electronic devices.

Modified Feature Mode Decomposition Guided by Spectral Structure Information for Machinery Fault Diagnosis

Modified Feature Mode Decomposition Guided by Spectral Structure Information for Machinery Fault Diagnosis

Variational multi-harmonic duality mode pursuit method for extracting repetitive transient components from vibration signals

The repetitive transient component (RTC) in vibration signals is a typical symptom reflecting the localised defects of rotating machinery, which can be mathematically defined by feature parameters, such as the inter-arrival interval, maximum amplitude, resonance frequency, and attenuation ratio. In particular, the inter-arrival interval and maximum amplitude reveal the location and severity of failure and provide critical information for condition monitoring and fault diagnosis. However, there are some shortcomings of applicability, anti-noise, and fidelity, which are unfavourable for pursuing fault-induced RTC and enabling reliable recognition of key feature parameters. Hence, a novel method called variational multi-harmonic duality mode pursuit (VMHDMP) is proposed to extract the RTC accurately. A multi-harmonic duality mode function (MHDMF) that comprehensively depicts the critical feature parameters of the RTC was first defined in the frequency domain. Based on the defined MHDMF, the pursuit of RTC is formulated as a frequency-domain variational optimisation problem with a local compact time duration constraint, which optimally searches for the mode with an explicit MHDMF from the complex-valued frequency signal. A strategy involving the mutual conversion of real- and complex-valued signals was incorporated into the solution algorithm to satisfy the decomposability of complex-valued frequency signals. Finally, an inverse Fourier transform is performed on the analytical form of the extracted MHDMF to obtain a faithful RTC. Essentially, VMHDMP provides an optimal ‘transient-pass’ inverse filter bank with compact passage only near the occurrence moment of each transient in RTC. Thus, the vital parameters closely related to the faults are fully retained, whereas the frequency-overlap noise and large-amplitude interference occurring at different moments can be effectively eliminated. Simulated and experimental examples were used to test the effectiveness of VMHDMP.

A Novel Lithium-Ion Battery Impedance Fast Measurement Method With Enhanced Excitation Signal

Online battery impedance acquisition is of increasing interest due to its valuable information for state estimation and fault diagnosis. The Pseudo Random Sequence (PRS) can significantly mitigate the required time cost, yet suffers from an insufficient Signal-to-Noise Ratio (SNR). Motivated by this, this paper proposes a novel series-connected Dual Pseudo Random Sequence (DPRS) for enhancing the signal power content in all frequency ranges, which notably improves the robustness of broadband impedance acquisition against interference. A Weighted Bilateral Impedance Filter (WBIF) is further carried out on a semi-logarithm scale to process the measurement data, which effectively compensates for the noise-induced biases and enables a nearly unbiased result. Experimental studies verify the proposed framework by comparing it with the conventional PRS method in terms of accuracy and robustness under various States of Charges (SOCs), ambient temperatures, and aging status of the batteries.

Singular component decomposition and its application in rolling bearing fault diagnosis

In the rolling bearing fault diagnosis based on vibration signal, the feature information always exists in the specific sideband of the signal. Singular value, as a mathematical quantity can represent the characteristic information of data, provides the theoretical support for extracting fault information from signals. Based on this, this paper proposes a new method called Singular Component Decomposition (SCD). The signal is divided into components by the signal spectrum trend. Through linear transformation of Hankel matrix constructed by each component, the corresponding singular values are calculated. Combined with the amplitude filtering characteristic and the time domain negative entropy index, the effective singular values are screened out. The reconstructed components retain complete fault information for fault diagnosis of rolling bearings. The simulation and experimental results show that the SCD can extract the effective information to the maximum extent to realize the fault diagnosis of the signal.

Joint alignment network preserving structural information for multimode process fault diagnosis

AbstractProduction conditions are complex and varied for a number of reasons. Models for defect diagnosis may perform worse as a result of the distributional mismatch between test data and training data. In order to diagnose process faults, it is crucial to take into account the fact that data exhibits varied distribution characteristics under various conditions. In the case of multiple operating conditions, the cross‐domain problem caused by different data distributions can degrade the performance of deep learning‐based fault diagnosis models. To overcome this challenge, a joint alignment network preserving structural information (JANSI) method is proposed. To extract richer and fine‐grained domain‐invariant features, the structural information preservation is proposed, which combines domain labels, category labels, and data distribution structures. To increase intra‐class compactness and inter‐class separability, class centre alignment is proposed. The effectiveness of the method on the cross‐domain unsupervised fault diagnosis problem is verified through three case studies.

Method of Virtual Sensor Design for Faulty Physical Sensor Replacement

The paper considers the problem of virtual sensor design for nonlinear dynamic systems with non-smooth nonlinearities described by continuous-time models for faulty physical sensor replacement. The main purpose of virtual sensors is generating the estimates of the unmeasured components of the considered system to provide additional information for effective control and fault diagnosis. Besides, virtual sensors can be used for faulty physical sensor replacement. The methods of virtual sensor design for solving this problem differ from standard procedure since information from faulty physical sensor does not use to design the virtual sensor replacing this sensor. It is assumed that to solve the problem, the system is equipped by diagnostic system allowing detecting faulty sensor. For every such a sensor, the virtual sensor generating estimate replacing the faulty sensor is designed. To solve the problem, so-called logic-dynamic approach is used which does not guarantee optimal solution but uses only methods of linear algebra to solve the problem for systems with non-smooth nonlinearities. This approach contains three steps. Initially, the nonlinear term is removed from system and linear model is designed. Then, a possibility to estimate the faulty sensor and to insert in the model the transformed nonlinear term is checked. Finally, stability of sensor is provided. The virtual sensor can be designed in identification canonical form or Jordan canonical form. The advantage of the first form is a standard procedure of the virtual sensor design while Jordan form allows obtaining simpler solution. The relations allowing designing the virtual sensor as in identification canonical as Jordan canonical form are derived.

Denoising graph neural network based hydraulic component fault diagnosis method

Accurately identifying faults in hydraulic components is essential for improving the reliability and safety of hydraulic systems. Since the hydraulic components of heavy-duty equipment usually work in complex and variable environments and are subject to interference from multiple sources of noise. As a result, fault diagnosis of hydraulic components is challenging, leading to the development of various fault diagnosis models. However, current hydraulic component fault diagnosis models have the following limitations: (1) only able to handle Euclidean space data and ignore the topological relationships within the signals that could provide additional useful information for fault diagnosis, (2) only diagnose faults using single-dimensional acceleration data and fail to incorporate more equipment working conditions. This study focuses on the hydraulic directional valve and proposes a fault diagnosis method based on denoising graph neural network. Firstly, computational fluid dynamics(CFD) is used to analyze the flow field of the spool and sleeve of the valve under different wear states and working conditions. It is found that different degrees of wear have similarities under different working conditions, and there is a regular correlation between the acceleration signal of the valve body and the fault mode under the same working condition. Therefore, acceleration and working condition information are selected as the raw information for fault diagnosis. Then, horizontal visibility graph(HVG) was utilized to transform acceleration signal to graph, and the graph nodes, which can incorporate multidimensional information, were set to be the equipment working conditions and acceleration values. However, the noise in the raw signal could seriously affect the quality of the HVG. To solve this problem, an adaptive threshold denoising algorithm was proposed to reduce the interference noise in the acceleration signal. The denoising threshold parameter was learned directly from the raw data. Finally, graph neural network(GNN) is applied to fault diagnosis of hydraulic components from non-Euclidean multidimensional information. Experimental results demonstrate the superior robustness and accuracy of this method in handling strong noise and multiple working conditions compared to several classical fault diagnosis methods. This research provides an effective approach for hydraulic component fault diagnosis.

VWM-DCRNN: A Method of Combining Signal Processing With Deep Learning for Fault Diagnosis in Single-Phase PWM Rectifier

Based on variational mode decomposition and dual model of convolutional recurrent neural network (VMD-DCRNN), a novel diagnosis method is proposed to discover the faults of IGBTs, diodes and series resonant circuit in single-phase pulse width modulation (PWM) AC-DC rectifier, an important part of traction power supply system of high-speed railway. By virtue of the combination of signal processing and deep learning schemes, the multi-scale feature information for fault diagnosis is extracted by regarding the input current of the rectifier and the voltage of dc side as the original signal. More clearly, VMD is adopted to decompose the original signal to a series of intrinsic mode function components (IMFs). Then, the submodel of current is designed to detect the fault types of IGBTs and diodes. Meanwhile, the submodel of voltage is established to identify the fault types of series resonant circuit. These IMFs are fed into DCRNN consisting of the two submodels for training and testing. Finally, the features extracted from the two submodels are integrated and converted to label space to complete fault diagnosis. Experimental results show that the proposed VMD-DCRNN algorithm is superior to the single-channel models, reaching 96.27%.

RFR-GA-BLS: A Feature Selection and Parameter Optimization Method for Fault Diagnosis of Rolling Bearing Using Infrared Images

To overcome the problems of low machine learning fault diagnosis rate and long consumption time of deep learning in rolling bearing fault diagnosis, an RFR-GA-BLS model is proposed. The model is validated by infrared images of rolling bearings to find the most representative features, the most suitable parameters and the best diagnostic rate. Based on the pre-processed infrared thermal images of the faulty bearing, 72 second-order statistical features were obtained as information for fault diagnosis. RFR considered the robustness of the features, and new sequences were obtained. BLS was optimized by GA for fault diagnosis. New sequence features were added to the model sequentially, one at a time. After satisfying the model conditions, the most appropriate number of features was selected as the first 20. The search results for the number of feature nodes, the number of feature node windows and the number of enhancement nodes for the BLS were 24, 19 and 544, respectively, and the fault diagnosis rate of 98.8889% was achieved. According to a comparison with CFR-GA-BLS, BLS, PSO-BLS and Grdy-BLS, our proposed model is more advantageous in the search for the best performance. The fault diagnosis accuracy is higher compared to SVM and RF. The speed of our proposed model is 207 times faster than 1DCNN and 10,147 times faster than 2DCNN.

A Two-Stage Optimization Framework for Fast Lithium-Ion Battery Impedance Measurement

Online monitoring of battery impedance is of increasing interest as it brings rich information for battery state estimation and fault diagnosis, however, the existing techniques are limited by high implementation costs and poor accuracy. This letter addresses the challenges by proposing a novel two-stage optimization framework to form a binary sequence with the optimal power proportion in specified harmonics for fast battery impedance measurements. The Optimized Binary Sequence (OBS) with finite amplitude levels can greatly simplify the implementation, and also the optimized power spectrum has superb accuracy to obtain the frequency responses. Experimental studies confirm the superiority of the proposed method in comparison with three existing techniques under various circumstances.

Research on the distribution mechanism of fault alarm information for power grid fault diagnosis

Power grid fault diagnosis aims to locate faulty components quickly and accurately. The alarm information generated by the SCADA system of the grid is the data basis for power grid fault diagnosis. The existing research lacks mechanical analysis of the alarm information, so the models and algorithms are difficult to be used in engineering applications. The established power grid fault diagnosis models are somewhat different from the actual grid, and the alarm information timing characteristics are difficult to express. In this paper, we analyze the correlation and characteristics of the data related to the number of voltages, number of stations, and duration of faults for each period, and analyze the spatial and temporal distribution mechanism of the alarm messages in conjunction with the characteristics of the equipment that issued the alarm messages during the fault events. It also compares the differences in the distribution patterns of the alarm information when the data sources are different. The research can lay the foundation for establishing fault diagnosis models and provide theoretical support for model design and algorithm research.

A Text-Oriented Fault Diagnosis Method for Electromechanical Device Based on Belief Rule Base

At present, quantitative data is often used for fault diagnosis of electromechanical devices, while qualitative data in the form of text is rarely used. In order to integrate qualitative data in the form of text and quantitative data in the fault diagnosis of an electromechanical device, a text-oriented fault diagnosis method based on belief rule base (BRB) is proposed in this paper. Specifically, the key information of fault diagnosis is extracted from the text through natural language processing (NLP) and then converted into belief rules. Then, a rule supplement method is adopted to add the extracted belief rules to the BRB for the completion of the BRB construction. This method applies qualitative data in the form of text to the process of BRB construction, which is a new attempt at the BRB construction method. It not only solves the problem that BRB cannot use qualitative data in text form but also improves the modeling accuracy and data comprehensive processing ability of BRB. To verify the effectiveness of the algorithm, we designed an experiment of asynchronous motor fault diagnosis in the case study. The experimental result shows that the proposed method can use qualitative data in text form to construct BRB and effectively diagnose faults of asynchronous motors. The MSE of the proposed method is 0.0451, which is better than that of traditional BRB (0.1461), BP (0.0613), and SVR (0.0974) under the same experimental conditions.