Incipient Fault Detection Research Articles

A new framework based on geometric partition L-Kurtosis indicator for bearing condition monitoring and incipient fault detection

Health indicators (HIs) are greatly significant in sustaining the stable operation of machines and guiding the implementation of predictable maintenance. Traditional HIs such as kurtosis, negentropy, Gini index manifest some concessions in certain scenarios. In this study, we propose a new geometric partition L-Kurtosis (GLK) indicator for bearing condition monitoring leveraging its splendid fault characterization, and it is proved to be advantageous in perceiving bearing incipient fault occurrence and outlining sufficiently bearing full-life degradation trend through substantial experiments with run-to-failure datasets. Integrated with empirical wavelet transform, a GLKgram is constructed to enable the application of the GLK indicator in bearing incipient fault detection. The experimental results deliver that the GLKgram excels at recognizing the fault repetitive transients and resisting the interfering large random shocks and strong noise, further manifesting the superiority of the GLK indicator in identifying fault features.

Investigations on improved Box-Cox sparsity measures for machine condition monitoring.

Sparsity measures are commonly utilized as health indicators for machine condition monitoring. Recently, with the assistance of Box-Cox transformation, kurtosis and negative entropy have been smoothly extended to form Box-Cox sparsity measures (BCSMs). However, traditional BCSMs do not generate sparsity measures that outperform negative entropy, which means it is meaningless to some extent. Therefore, in this paper, traditional BCSMs are further extended to develop more robust sparsity measures. Firstly, inspired by the limited weight range of the Gini index, the traditional BCSMs are extended to the case of λ<0 by a two-parameter Box-Cox transformation. Then, by examining the decomposition forms of L2/L1 norm and Hoyer measure, the advantage of directly applying the Box-Cox transformation to the sparsity measure is discovered. Thus, the improved BCSMs (IBCSMs) are naturally proposed by performing the classical Box-Cox transformation on the BCSMs with λ≥-1. Subsequently, three key properties of the proposed sparsity measures are analyzed through three numerical experiments. Finally, the proposed sparsity measures are deployed as health indicators to characterize the degradation process of three bearings. Numerical and experimental results demonstrate the salient advantages of the proposed IBCSMs in incipient fault detection.

Incipient Fault Detection and Recognition of China Railway High-Speed (CRH) Suspension System Based on Probabilistic Relevant Principal Component Analysis (PRPCA) and Support Vector Machine (SVM)

As a crucial component of CRH (China Railway High-speed) trains, the safety and stability of the suspension system are of paramount importance to the overall vehicle system. Based on the framework of probabilistic relevant principal component analysis (PRPCA), this paper proposes a novel method for incipient fault diagnosis in the CRH suspension system using PRPCA and support vector machine (SVM). Firstly, simulation data containing multiple types of fault information are obtained from the Simpack2018.1-Matlab2016a/Simulink co-simulation platform. Secondly, the nonlinear PRPCA approach, based on the Wasserstein distance, is employed for fault detection and data preprocessing in the suspension system. Furthermore, SVM is used for fault recognition, and the F1-Measure index is utilized for a comprehensive evaluation to assess the fault diagnosis performance more intuitively. Finally, based on the comparison results with traditional principal component analysis (PCA) and SVM-based methods, the proposed incipient fault diagnosis method demonstrates superior efficiency in fault detection and recognition. However, the proposed method is not very sensitive to sensor faults, and the performance of sensor fault diagnosis needs to be further improved in subsequent research.

Incipient fault detection for the spindle bearing of a cement grinding machine based on vibrational resonance

Abstract The spindle bearing of a circulation fan is an important component of a cement grinding machine. In addition to the faults on the spindle bearing, impeller wear and ash accumulation may cause dynamic unbalance and complex vibration interference as noise, which decreases the accuracy of fault detection based on vibrational signals and traditional signal processing-based methods at the early stage of a bearing fault. To address this issue, this paper presents a new fault detection method for the spindle bearing by utilizing extra injected noise and vibrational resonance. To enhance the fault signature and resonance performance, the nonlinear system of the traditional vibrational resonance is replaced by a new hybrid steady-state system, and the underdamped term is considered in the new system. The proposed system provides more possibilities to achieve resonance by adjusting the system parameters and overcomes the limitations of output saturation caused by the classical bistable system. The proposed method is validated by analyzing the collected vibration signals from a spindle bearing of a circulation fan in practice and is compared with other noise-elimination fault detection methods. The results demonstrate the effectiveness and superior performance of the proposed method.

Unsupervised motor incipient fault detection using lightweight network and orthogonal low-rank embedding

Abstract Electrical motor is a key component in industrial systems. Detecting incipient fault of motor is critical to system reliability. However, the characteristics of faults and normal stages are difficult to distinguish, Meanwhile, the lightweight requirements of the model and the imbalance of the samples also make incipient fault detection challenged. To address these problems, this paper proposes an unsupervised fault detection method combines lightweight network and orthogonal low-rank embedding (OLE). The raw signals are firstly transformed into time-frequency images and fed into lightweight convolution network for feature extraction. Then, the features are clustered into orthogonal subspaces to enhance inter-class separability. Finally, a detection module based on distance metric is designed to identify the incipient fault of motor. The effectiveness of the proposed method is validated on four industrial motor dataset and compared with other methods.

Incipient fault detection and isolation with Cauchy–Schwarz divergence: A probabilistic approach

To monitor the dynamics and non-stationarity inherent in industrial processes, we propose a novel incipient fault detection and isolation scheme grounded in a probabilistic perspective, using the Cauchy–Schwarz (CS) divergence. Our innovation lies in the utilization of marginal CS divergence for incipient fault detection and the conditional CS divergence for fault isolation. This approach neither require prior parametric assumptions about the underlying data distribution nor depend on historical fault data, while simultaneously providing explanatory diagnostics. Beyond this, we develop a change point detection-base diagnosis technique for practical engineering applications. This online process monitoring technique guarantees timely intervention to uphold process stability and safety. We demonstrate the compelling performance, higher detection rate and lower alarm rate, of the CS divergence over prevalent divergence-based approaches, such as Kullback–Leibler divergence, Wasserstein distance and Mahalanobis distance. We also illustrate the explanatory insights offered by conditional CS divergence in fault isolation on synthetic data, benchmarks of continuous stirred-tank reactor process and continuous stirred-tank heater process and even a real-world continuous catalytic reforming process. Code of this CS divergence based fault detection and isolation is available at https://github.com/Feiya-Lv/Incipient-Fault-Detection-and-Isolation-with-Cauchy--Schwarz-Divergence.

Robust optimized weights spectrum: Enhanced interpretable fault feature extraction method by solving frequency fluctuation problem

Machine condition monitoring (MCM) plays a pivotal role in ensuring the reliability, safety, and efficiency of a production and operation system. Fault feature extraction (FFE), as an important step within MCM, aims to filter out interference components (ICs) and extract fault components (FCs) from raw signals. Consequently, it facilitates incipient fault detection and performance degradation assessment. A recently proposed optimized weights spectrum (OWS) provided a data-driven FFE methodology with a strong theoretical foundation, but it was found that the OWS is highly sensitive to a frequency fluctuation problem caused by rotational speed fluctuations and sensor errors, which may result in fake fault signatures. Therefore, a robust optimized weights spectrum (ROWS) is proposed in this paper to solve the frequency fluctuation problem. The ROWS is inspired by an intuitive idea that employs functions with adaptive frequency bandwidths instead of isolated spectral lines to represent frequency components, thus mitigating the impact of the frequency fluctuation problem. To adaptively determine the optimal parameters for these functions and estimate the corresponding ROWS, an alternative optimization strategy is adopted. Then, the proposed ROWS can be obtained when convergence conditions of this strategy are satisfied. Finally, the effectiveness and superiority of the proposed ROWS are verified by three real-world cases, i.e., the proposed ROWS can solve the frequency fluctuation problem and provide robust results for interpretable FFE.

Relation between fault characteristic frequencies and local interpretability shapley additive explanations for continuous machine health monitoring

Recently, the Shapley additive explanations models have been extensively studied to enhance explainability of artificial intelligence algorithms, while most of them simply use Shapley additive explanations to rank or measure the importance of different features. In this study, a novel methodology that studies the relation between fault characteristic frequencies and Shapley values generated by local interpretability Shapley additive explanations for machine health monitoring is proposed. Firstly, a simulation model is introduced to generate vibration signals at different health conditions and their spectral amplitudes transformed from Fourier transform are used to investigate the relationship between fault characteristic frequencies and local interpretability Shapley values. It is interestingly found that Shapley values can be used to locate fault characteristic frequencies. Moreover, most of them have negative values in a normal stage and have positive values in an abnormal stage. Based on this finding and Shapley additive explanations, a health indicator construction methodology is proposed to continuously monitor incipient machine faults. Subsequently, an automatic signal filtering method is proposed to remove and eliminate burrs and noise in Shapley values so that fault characteristic frequencies can be clearly revealed by Shapley values for physical fault diagnosis. Two run-to-failure cases are conducted to demonstrate the effectiveness of the proposed methodology and then the superiority of this study is demonstrated by comparing with existing methods for health indicator construction and fault diagnosis, including sparsity parameters, Hjorth parameters, and fast Kurtogram. Comparison results show that the proposed health indicator is more sensitive to the time of incipient fault initiation and interpretable fault diagnosis based on Shapley values has a robust performance. This study first sheds a light on the relationship between fault characteristic frequencies and Shapley values under the scenario of continuous machine health monitoring and seamlessly guides applicants to realize Shapley additive explanations based incipient fault detection and diagnosis.

Incipient fault detection for dynamic processes with canonical variate residual statistics analysis

In modern complex industrial operations, timely fault detection is imperative. While statistical process monitoring is widely used in practice, conventional approaches are usually insensitive to incipient faults (IFs) whose magnitudes are not obvious. To this end, an innovative approach is presented for IF detection in dynamic processes. To begin with, canonical variate residuals (CVRs) are generated by using the canonical variate dissimilarity analysis (CVDA) algorithm. The next step involves calculating statistics for the CVRs and arranging a corresponding statistic matrix. Afterward, the Mahalanobis distance index is constructed for fault detection purpose. The main reasons that this developed approach possesses high sensitivity to IFs in dynamic processes lie in the utilization of CVDA and the idea of monitoring extracted statistics rather than original residuals. Finally, its effectiveness and merits are demonstrated via a numerical example and a benchmark process.

Incipient fault detection based on dense feature ensemble net

With modern industrial processes becoming more and more complex, the occurrence of faults may cause unmitigated disaster. Therefore, incipient fault detection is very important and has attracted increasing attention Recently, a feature ensemble net (FENet) has been proposed based on a deep feature ensemble framework, which can detect partial notorious faults (such as faults 3, 9 and 15 in the Tennessee Eastman process (TEP)). However, the feature transformer layers in FENet can only mine the process information from the upper layer’s output rather than all features from previous layers. Therefore, a novel dense feature ensemble net (DenseFENet) is proposed to deeply extract the features of data for incipient fault detection in industrial process. DenseFENet firstly constructs base detectors with capabilities for extracting nonlinear features, such as one-class support vector machine, isolation forest, one-class back propagation neural network, one-class long short term memory network and one-class temporal convolutional network, etc. Then, a dense net structure with short paths between non-adjacent feature transformer layers is developed, improving the reutilization capability for shallow knowledge. In addition, a numerical simulation and TEP are adopted to demonstrate the performance of the proposed method. The fault detection rate of DenseFENet has improved by 5.05% and 3.27% correspondingly, revealing the superiority of this approach.

T-S Fuzzy Data-Driven ToMFIR With Application to Incipient Fault Detection and Isolation for High-Speed Rail Vehicle Suspension Systems

T-S Fuzzy Data-Driven ToMFIR With Application to Incipient Fault Detection and Isolation for High-Speed Rail Vehicle Suspension Systems

Detection of seeded fault and growing fault in rotating machinery under different speed scenarios using adaptive resonance-based signal sparse decomposition

Detection of seeded faults (SFs) in machinery elements such as gears and bearing has been widely explored, whereas there is limited research considering the growing fault (GF). The literature indexes separate techniques for diagnosing SFs and GFs, because of their nature. No technique yet reported is capable to detect both types of faults. This work proposes a novel adaptive approach and an uncertainty index (UI) to detect both SF and GF. In the proposed approach, the signal is pre-processed by adaptive resonance-based signal sparse decomposition to get low-resonance components (LRCs) and high-resonance components across different decomposition levels. Using the UI, the fault prone LRC is identified based on impulsiveness and complexity content. Afterwards the empirical mode decomposition (EMD) is applied to extract the intrinsic mode functions. The Fast Fourier Transform (FFT)s of the modes extracted from EMD reflected the characteristic frequencies and increased amplitude of sidebands. Lastly, the statistical analysis of the FFTs was performed using the sideband energy ratio to validate the state of faults. The proposed approach when examined for both SFs in gears and GFs in bearings at different conditions, yielded satisfactory results. The proposed approach performance was found to be at least 15% better for incipient fault detection and more than 50% for severe fault conditions.

Incipient fault detection and isolation for dynamic processes with slow feature statistics analysis

Incipient fault detection and isolation is crucial to maintain a high-efficiency operational state for modern complicated manufacturing processes. Nonetheless, many conventional data-driven methods are insensitive to incipient faults with tiny magnitudes. To address this issue, a novel process monitoring method called slow feature statistics analysis (SFSA) is proposed. Slowly changing features containing significant information are first extracted, and the sliding window technique is applied to acquire their statistical information. Afterward, two detection indices are constructed for fault detection. After successful detection, the reconstruction-based grouped contribution for the statistics of slowly changing features is calculated, which is used to locate abnormal slow features. Then, the abnormal variable is identified by implementing a linear mapping strategy. The evaluation for SFSA is conducted by using a numerical example and a benchmark process. The results indicate that SFSA performs better than several other methods in terms of detection sensitivity and isolation accuracy.

A Fault-Targeted Gated Recurrent Unit-Canonical Correlation Analysis Method for Incipient Fault Detection

A Fault-Targeted Gated Recurrent Unit-Canonical Correlation Analysis Method for Incipient Fault Detection

A transferable diagnosis method with incipient fault detection for a digital twin of wind turbine

Accurate and transferable fault diagnosis methods have a critical role in constructing a digital twin (DT) of wind turbine (WT). These methods can be utilized to predict premature failures and to maintain a stable power supply. However, the infeasibility of obtaining enough degradation data with various failure mechanisms is one of major problems for WTs. This paper proposes a fault detection and diagnosis method for WTs using tools, including average integrated power spectral density (AIPSD), one-class support vector machine (OCSVM), and neural architecture optimization (NAO)-CapsNet, based on limited data. Moreover, a WT DT framework is designed for real-time condition monitoring of the WT during its lifecycle. Datasets collected from real-world WT gearboxes and pitch bearings are selected to verify the performance of the proposed method. The experimental results show that the developed method can not only trace when the fault occurred, but also obtain better diagnosis accuracy compared with other state-of-the-art methods. In addition, the proposed NAO-CapsNet trained by dataset from two WTs can be successfully transferred and applied to DT-based condition monitoring of another WT.

Optimal weight impulse extraction: New impulse extraction methodology for incipient gearbox condition monitoring

Gear faults in a transmission system generally cause impulse components in vibration signals, which is a crucial symbol for gearbox fault diagnosis. However, their related signals are often interfered or even submerged by the noisy meshing components (NMC) of gearboxes in degradation, which introduces challenges for incipient fault detection and condition monitoring. Commonly employed deconvolution-based methods attempt to design a filter to extract impulse components. However, these methods fail to address the interference issue of the NMC on deconvolution process. To overcome this limitation, this paper proposes an optimal weight impulse extraction (OWIE) methodology to suppress the NMC and highlight impulse component in vibration signal. Different from deconvolution-based methods, the proposed method locates impulses adaptively driven by waveform itself. A non-impulse part is suppressed through point-to-point removal, while impulse components are highlighted by subtracting a weighted signal from a raw signal. An iterative procedure is utilized to solve the optimal weight sequence by maximizing the kurtosis of an impulse signal. The effectiveness of the proposed OWIE is validated through a simulation case study and a run-to-failure experiment of gearboxes. Results demonstrate that the OWIE is sensitive to incipient faults and is suitable for the health condition monitoring of gearboxes.

Incipient Fault Detection-Based Active Protection of Photovoltaic DC Integration System

Incipient Fault Detection-Based Active Protection of Photovoltaic DC Integration System

Detection of Incipient Faults in Power Transformers using Fuzzy Logic and Decision Tree Models Based on Dissolved Gas Analysis

This paper proposes an integrated approach utilizing Fuzzy Logic and Decision Tree algorithms to diagnose early-stage faults in power transformers based on Dissolved Gas Analysis (DGA) test results of transformer insulation oil. Overcoming limitations in conventional methods such as Duval Triangle, Key Gas Analysis, Rogers Ratio, IEC Ratio, and Doernenburg Ratio, our Fuzzy Logic and Decision Tree models address issues like inaccurate diagnosis, inconsistent diagnosis, lack of decisions or out-of-code results, and time-intensive manual calculations for large DGA datasets. The Decision Tree algorithm, a machine learning technique is applied to categorize faults into thermal and electrical types. Trained with over 300 DGA samples from transformers with known faults, the models exhibit robust performance during testing with different datasets. Notably, the Duval Triangle decision tree model attains the highest accuracy among the ten developed models, achieving a 98% accuracy rate when tested with 50 samples with known faults. Moreover, Decision Tree models for KGA, Doernenburg, Rogers, and IEC also demonstrate substantial prediction accuracy at 92%, 86%, 92%, and 90% respectively underscoring the efficacy of artificial intelligence methods over traditional approaches.

Enhancing Incipient Fault Detection for Interface Converter Sensors through Signal Correlation Analysis

Incipient faults in interface converters can potentially lead to catastrophic failures. Detection of incipient faults contributes to proactive fault management and predictive maintenance, which effectively improves system reliability. In this paper, a detection method in correlation space is proposed to address this problem, which is based on the inherent feature of correlation changes when a fault occurs. First, a convolution process is used to weaken noise and highlight the correlation feature. Second, the current signals are transmitted to correlation space by using the Pearson correlation coefficient. Third, an accumulation and a reference compensation method are designed for enhancing the features and equalizing influence of reference adjustment. Finally, a fault detection rule is designed based on correlation features and fault excitation. Experiments on a hardware‐in‐loop(HIL) semi‐physical platform indicate that the proposed method outperforms three commonly used correlation analysis algorithms. © 2024 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC.

Distributed incipient fault detection with causality-based multi-perspective subblock partitioning for large-scale nonlinear processes

Detecting incipient faults is a crucial task in complex large-scale nonlinear industrial processes. However, conventional distributed frameworks encounter challenges in identifying appropriate subblock partitioning methods. To address this problem, we propose a novel distributed incipient fault detection method, called CMPSP-MKPCA, based on causal relationship multi-perspective subblock partitioning (CMPSP) and modified kernel principal component analysis (MKPCA), facilitating incipient fault detection in large-scale nonlinear processes. Firstly, considering the property of incipient faults to undergo limited propagation among process flow connections, we obtain fine-grainehe direction of potential incipient fault information propagation. Secondly, the design of multi-perspective detection blocks provides real-time selection of sub-block delineation results to adapt to the different needs of incipient faults during latency and transition periods. Thirdly, local detection is realized by combining the Modified KPCA method and Kullback-Leibler divergence (KLD), while the global detection statistics is obtained by Bayesian inference fusion. Simulation experiments based on the continuous stirred tank reactor (CSTR) system and the Tennessee Eastman (TE) system verify the effectiveness and superiority of the method.