Machine Fault Detection Research Articles

Vibration Image Representations for Fault Diagnosis of Rotating Machines: A Review

Rotating machine vibration signals typically represent a large collection of responses from various sources in a machine, along with some background noise. This makes it challenging to precisely utilise the collected vibration signals for machine fault diagnosis. Much of the research in this area has focused on computing certain features of the original vibration signal in the time domain, frequency domain, and time–frequency domain, which can sufficiently describe the signal in essence. Yet, computing useful features from noisy fault signals, including measurement errors, needs expert prior knowledge and human labour. The past two decades have seen rapid developments in the application of feature-learning or representation-learning techniques that can automatically learn representations of time series vibration datasets to address this problem. These include supervised learning techniques with known data classes and unsupervised learning or clustering techniques with data classes or class boundaries that are not obtainable. More recent developments in the field of computer vision have led to a renewed interest in transforming the 1D time series vibration signal into a 2D image, which can often offer discriminative descriptions of vibration signals. Several forms of features can be learned from the vibration images, including shape, colour, texture, pixel intensity, etc. Given its high performance in fault diagnosis, the image representation of vibration signals is receiving growing attention from researchers. In this paper, we review the works associated with vibration image representation-based fault detection and diagnosis for rotating machines in order to chart the progress in this field. We present the first comprehensive survey of this topic by summarising and categorising existing vibration image representation techniques based on their characteristics and the processing domain of the vibration signal. In addition, we also analyse the application of these techniques in rotating machine fault detection and classification. Finally, we briefly outline future research directions based on the reviewed works.

A Method for Statistical Processing of Magnetic Field Sensor Signals for Non-Invasive Condition Monitoring of Synchronous Generators.

Condition monitoring of synchronous generators through non-invasive methods is widely requested by maintenance teams for not interfering the machine operation. Among the techniques used, external magnetic field monitoring is a recent strategy with great potential for detecting incipient faults. In this context, this paper proposes the application of a simple strategy with low computational cost to process data of external magnetic field time derivative signals for the purposes of condition monitoring and fault detection in synchronous machines. The information of interest is extracted from changes in the magnetic signature of the synchronous generator, obtained from frequency spectra of monitored signals using induction magnetic field sensors. The process forms a set of time series that reflects constructive and operational characteristics of the machine. The Shewhart control chart method is applied for anomaly detection in these time series, allowing the detection of changes in the machine magnetic signature. This method is employed in an algorithm for continuous condition monitoring of synchronous generators, presenting as output a global change indicator for the multivariable problem associated with magnetic signature monitoring. Correlation matrices are used to improve the algorithm response, filtering series with similar variation patterns associated with detected events. The proposed method is validated through tests on an experimental bench that allows the controlled imposition of faults in a synchronous generator. The proposed global change indicator allows the automatic detection of stator and rotor faults with the machine synchronized with the commercial power grid. The proposed methodology is also applied on data obtained from an equipment installed in a 305 MVA synchronous generator of a hydroelectric power plant where the evolution of an incipient fault, i.e., a mechanical vibration fault, has been detected.

Deep Cellular Recurrent Network for Efficient Analysis of Time-Series Data With Spatial Information.

Efficient processing of large-scale time-series data is an intricate problem in machine learning. Conventional sensor signal processing pipelines with hand-engineered feature extraction often involve huge computational costs with high dimensional data. Deep recurrent neural networks have shown promise in automated feature learning for improved time-series processing. However, generic deep recurrent models grow in scale and depth with the increased complexity of the data. This is particularly challenging in presence of high dimensional data with temporal and spatial characteristics. Consequently, this work proposes a novel deep cellular recurrent neural network (DCRNN) architecture to efficiently process complex multidimensional time-series data with spatial information. The cellular recurrent architecture in the proposed model allows for location-aware synchronous processing of time-series data from spatially distributed sensor signal sources. Extensive trainable parameter sharing due to cellularity in the proposed architecture ensures efficiency in the use of recurrent processing units with high-dimensional inputs. This study also investigates the versatility of the proposed DCRNN model for the classification of multiclass time-series data from different application domains. Consequently, the proposed DCRNN architecture is evaluated using two time-series data sets: a multichannel scalp electroencephalogram (EEG) data set for seizure detection, and a machine fault detection data set obtained in-house. The results suggest that the proposed architecture achieves state-of-the-art performance while utilizing substantially less trainable parameters when compared to comparable methods in the literature.

Remote Diagnosis and Detection Technology for Electrical Control of Intelligent Manufacturing CNC Machine Tools

An intelligent manufacturing environment employs internet-based communication and monitoring technologies for fault detection, diagnosis, and monitoring of industrial machines. The monitoring and fault detection are performed remotely without human intervention that predicts faults and ensures specific operational control. This article introduces a rational fault diagnosis process (RFDP) best suited for remote fault detection and diagnosis of CNC machine tools. The proposed process monitors different operational segments of the machine and extracts related data to validate its performance. The interconnection between the segments and fault impact are identified using the transfer learning process. The previously identified faults are used in the state training process to improve detection and diagnosis accuracy. Depending on the operational control continuity, the performance is assessed post the fault diagnosis. The learning paradigm is trained using the machine’s efficiency and rational data processing to predict the transfer states’ faults. The transfer states are modulated based on the efficiency and minimum-maximum control recommended for the CNC machine. This process’s performance is validated using detection accuracy, diagnosis recommendation, downtime, data processing rate, and processing time. From the experimental analysis, it is seen that for the varying data extraction rates, the proposed process improves detection accuracy by 10.14%, diagnosis recommendation by 8.58% and data processing rate by 7.95%, reducing the downtime by 8.85%, and processing by 11.24%.

Best Time Domain Features for Early Detection of Faults in Rotary Machines Using RAT and ANN

Background Bearing failure, the most frequent failure mode in rotating machinery is the typical mechanical fault. Such a failure might result in substantial financial losses at the workplace. One of the approaches made possible by other signal processing techniques is the early identification of various faults in rotating machinery; including bearing failures, misalignment, and others.PurposeThis fault is associated with many features used to diagnose different faults; thus, the Diagnostic Features (DF) is estimated at limited cyclic frequencies that refer to machine faults.MethodTwo methods are used to extract the DF. The first one depends on time-domain features. The second is based on an advanced representation of the frequency domain, which depends on spectral coherence (SCoh) data over the spectral frequency domain using a center frequency and frequency range determined by a 1/3 binary tree structure. The calculated DFs are represented by a 2D map against the center frequency and frequency resolution. The maps from different fault features are collected to form the diagnostic patterns. The best characteristics connected to these various flaws can be found using statistical techniques like reverse arrangement tests (RAT). Artificial neural networks (ANN) may be trained and auto-diagnosed using the results from the best characteristics.ResultsUsing RAT is considered very important to summarize features. This method is given good results in training and diagnosis.ConclusionsAdditionally, ANN and RAT provide a detection result of 100% based on the description of the machine's operating situation, whether it functioned commonly or incorrectly.

Diagnostic of Combined Mechanical and Electrical Faults in ASD-Powered Induction Motor Using MODWT and a Lightweight 1-D CNN

The early fault detection in the rotary electrical machines,such as induction motors (IMs), has been growing in modern industry. IMs have been widely used in industrial applications due to its easy installation, reliability, and low cost. However, the increasing usage of IMs also increases the need for timely maintenance in order to ensure their operation and a longer service life. This article proposes a new diagnosis methodology based on maximal overlap discrete wavelet transform and a lightweight 1-D convolutional neural network (CNN) architecture, in order to detect mechanical and electrical faults and their combination, in adjustable speed drive (ASD)-powered IMs. Specifically, single and combined faults were studied from the next: Outer raceway bearing (mechanical), turn-to-turn short-circuit, and phase-to-ground short circuit (electrical). The presented study was developed using current signals acquired from stators of IMs of 1 hp. The current signals are measured at powered conditions introduced by a power grid with a constant frequency at 60 Hz, and an ASD at three different frequencies. The proposed diagnostic methodology reaches more than 99% of accuracy.

Uncertainty Informed Anomaly Scores with Deep Learning: Robust Fault Detection with Limited Data

Quantifying the predictive uncertainty of a model is an important ingredient in data-driven decision making. Uncertainty quantification has been gaining interest especially for deep learning models, which are often hard to justify or explain. Various techniques for deep learning based uncertainty estimates have been developed primarily for image classification and segmentation, but also for regression and forecasting tasks. Uncertainty quantification for anomaly detection tasks is still rather limited for image data and has not yet been demonstrated for machine fault detection in PHM applications. In this paper we suggest an approach to derive an uncertaintyinformed anomaly score for regression models trained with normal data only. The score is derived using a deep ensemble of probabilistic neural networks for uncertainty quantification. Using an example of wind-turbine fault detection, we demonstrate the superiority of the uncertainty-informed anomaly score over the conventional score. The advantage is particularly clear in an ”out-of-distribution” scenario, in which the model is trained with limited data which does not represent all normal regimes that are observed during model deployment.

A Fault Detection Method for Electrohydraulic Switch Machine Based on Oil-Pressure-Signal-Sectionalized Feature Extraction

A turnout switch machine is key equipment in a railway, and its fault condition has an enormous impact on the safety of train operation. Electrohydraulic switch machines are increasingly used in high-speed railways, and how to extract effective fault features from their working condition monitoring signal is a difficult problem. This paper focuses on the sectionalized feature extraction method of the oil pressure signal of the electrohydraulic switch machine and realizes the fault detection of the switch machine based on this method. First, the oil pressure signal is divided into three stages according to the working principle and action process of the switch machine, and multiple features of each stage are extracted. Then the max-relevance and min-redundancy (mRMR) algorithm is applied to select the effective features. Finally, the mini batch k-means method is used to achieve unsupervised fault diagnosis. Through experimental verification, this method can not only derive the best sectionalization mode and feature types of the oil pressure signal, but also achieve the fault diagnosis and the prediction of the status of the electrohydraulic switch machine.

A linear time method for the detection of collective and point anomalies

AbstractThe challenge of efficiently identifying anomalies in data sequences is an important statistical problem that now arises in many applications. Although there has been substantial work aimed at making statistical analyses robust to outliers, or point anomalies, there has been much less work on detecting anomalous segments, or collective anomalies, particularly in those settings where point anomalies might also occur. In this article, we introduce collective and point anomalies (CAPA), a computationally efficient approach that is suitable when collective anomalies are characterized by either a change in mean, variance, or both, and distinguishes them from point anomalies. Empirical results show that CAPA has close to linear computational cost as well as being more accurate at detecting and locating collective anomalies than other approaches. We demonstrate the utility of CAPA through its ability to detect exoplanets from light curve data from the Kepler telescope and its capacity to detect machine faults from temperature data.

Application of Differential Magnetic Field Measurement (DMFM method) in winding fault detection of AC rotating machines as part of expert monitoring systems

Stator and rotor winding damages in rotating machines are result of electrical, mechanical, and thermal stress. Online magnetic field monitoring via permanently installed measuring coils inside air gap is a well-established methodology which enables winding fault detection. The paper deals with a new method for detection of stator and rotor winding inter-turn short circuits of synchronous machines and slip rings induction machines, as well as rupture of rotor bars and cage ring of induction machines. The method novelty is based on differential measurement of magnetic field by using two serial connected measuring coils. They are installed on the places (stator or rotor teeth) in the machine which have, by absolute value, equal magnetic vector potential. The distance between the measuring coils is n·tp, where tp is a pole pitch, and n = 1, 2, 3, 4,... is a multiple of the pole pitch. Measuring the coil-induced voltage enables us to detect stator and rotor winding faults, which means that measured voltage is approximately zero without fault and increases in the presence of fault. Analysis of the measuring signal allows us to detect and locate fault. With this new method it is possible with high sensitivity to determine winding fault, which enables more reliable fault detection. For example, in comparison with the motor current signature analysis method (the most widely used method for motor faults detection), this new method gives 200 times higher sensitivity to fault occurrence. Also, by using the DMFM method, faults can be detected in the time domain and there is no need for spectral or other complex signal analysis. This is very important because the measuring equipment used for machine fault detection can be simple and more economically acceptable. The DMFM method enables fault detection for even small machines with small expense in a very effective way. The only downside of the DMFM method is the fact that machine should be disassembled in order to install measuring coils. This problem is solved during the machine overhaul or during the manufacturing of the machine, when sensors can be easily implemented in the machine. For machines with large air gap, measuring coils can be installed without a machine disassembly. For the purpose of the method testing, numerous finite-element (FE) simulations on the 2- and 3-D machine models have been carried out to verify the method. Powerful numerical tools generate realistic results with properly selected starting and boundary conditions. By FEM models, actual machines with embedded measuring coils where created and simulated. The voltage induced inside the measuring coils is calculated for different machine states, load point and with and without a fault (broken rotor bar or inter-turn short circuit). Also, this method was experimentally validated via series of laboratory tests performed on the real electric machines specially designed for fault study (broken rotor bars, broken ring and inter-turn short circuits in a stator and rotor winding). Additionally, this method is applied on more than 20 real machines in industry. Due to the large amount of measured data, in this paper, it will be presented only one measurement performed on an induction motor on which we have detected one broken rotor bar. The thickness of the measuring coil designed in the printed circuit board technique is 0.3 mm. The number of turns is from 3 to 10. This new method and performed FEM calculations together with the experimental measurements improve fault detection portfolio knowledge that can be used in monitoring and diagnostics of rotating machines. Furthermore, this patent-pending method is already implemented in three innovative products placed on market (expert monitoring systems), so this method is fully confirmed in practice.

Parabolic tapering piezoelectric rotational energy harvester: Numerical analysis with experimental validation

Inadequate host beam design leads to the inefficiency of any piezoelectric vibration energy harvester. This article presents a parabolic tapering width piezoelectric energy harvester under rotational excitation. The Euler-Bernoulli theory, Hamilton's principle, and Galerkin's approach are used for mathematical modeling to investigate the frequency characteristics and open-circuit voltage response. Finally, experiments are conducted for validation. Compared to earlier reported designs, the proposed harvester produced high voltage-per-total mass (43.618 V/Kg) and performance amplification factor (4.18), besides the advantageous low natural frequency. This research’s findings can be used in real-time health monitoring of rotating machine components and fault detection mechanisms.

Kernel ridge regression-based chirplet transform for non-stationary signal analysis and its application in machine fault detection under varying speed conditions

The vibration signals of variable speed rotating machines are non-stationary. Time-frequency analysis (TFA) can effectively analyze non-stationary signals in time–frequency (TF) plane and polynomial chirplet transform (PCT) is one of widely adopted TFA methods. In PCT, a vital step is to approximate the instantaneous frequency (IF) of signals through polynomial approximation. However, the solution of polynomial approximation is easily affected by noise or disturbance, which greatly limits the ability of PCT to analyze noisy signals. To solve this issue, kernel ridge regression-based chirplet transform (KRR-CT) is developed to precisely characterize the TF features of non-stationary signals and produce an energy concentrated TF plane. In the KRR-CT, even in the presence of severe noise, a stable solution can be obtained in the approximation step implemented through KRR. Moreover, KRR-CT based iterative algorithm is further constructed for the analysis of common multi-component signals. The efficacy of proposed KRR-CT based iterative algorithm is confirmed by synthetic and real signals, and the feasibility of applying it for machine fault detection is illustrated. The results indicate that the proposed methods can produce a more energy concentrated TF plane and provide more precise IF information for machine fault detection under varying speed conditions than four comparison methods.

Optical FBG-T Based Fault Detection Technique for EV Induction Machines

Electric vehicles (EV) represent a key technology to achieve a low-carbon transportation objective, whist induction motors are one of the promising topologies. The reliability of these machines is crucial to minimize the downtime, cost and unwanted human lives. Although several techniques are utilized in the condition monitoring and fault detection of electrical machines, there is still no single technique that provides an all-round solution to fault detection in these machines and thus hybrid techniques are used widely. This paper presents a novel non-invasive optical fiber technique in condition monitoring of induction machines and in the process detecting inter-turn short circuit faults. Owing to optical fiber’s immunity to magnetic flux, a composite FBG-T sensor formed by bonding a giant magnetostrictive transducer, Terfenol-D, onto a fiber Bragg grating is utilized to sense machines’ stray flux as a signature to determine the internal winding condition of the machines. A tri-axial auto datalogging flux meter was used to obtain the stray magnetic flux and test results obtained via LabView were analyzed in MatLab. Experimental and numerical results agree with each other and how that the FBG-T sensor accurately and reliably detected the short-circuit faults. Bragg shifts observed under short-circuit faults were in 100s of picometre range under various operating frequencies compared to the mid-10s of picometre obtained under healthy machine condition. These provide much promise for future EVs.

Short Circuit and Broken Rotor Faults Severity Discrimination in Induction Machines Using Non-invasive Optical Fiber Technology

Multiple techniques continue to be simultaneously utilized in the condition monitoring and fault detection of electric machines, as there is still no single technique that provides an all-round solution to fault finding in these machines. Having various machine fault-detection techniques is useful in allowing the ability to combine two or more in a manner that will provide a more comprehensive application-dependent condition-monitoring solution; especially, given the increasing role these machines are expected to play in man’s transition to a more sustainable environment, where many more electric machines will be required. This paper presents a novel non-invasive optical fiber using a stray flux technique for the condition monitoring and fault detection of induction machines. A giant magnetostrictive transducer, made of terfenol-D, was bonded onto a fiber Bragg grating, to form a composite FBG-T sensor, which utilizes the machines’ stray flux to determine the internal condition of the machine. Three machine conditions were investigated: healthy, broken rotor, and short circuit inter-turn fault. A tri-axial auto-data-logging flux meter was used to obtain stray magnetic flux measurements, and the numerical results obtained with LabView were analyzed in MATLAB. The optimal positioning and sensitivity of the FBG-T sensor were found to be transverse and 19.3810 pm/μT, respectively. The experimental results showed that the FBG-T sensor accurately distinguished each of the three machine conditions using a different order of magnitude of Bragg wavelength shifts, with the most severe fault reaching wavelength shifts of hundreds of picometres (pm) compared to the healthy and broken rotor conditions, which were in the low-to-mid-hundred and high-hundred picometre (pm) range, respectively. A fast Fourier transform (FFT) analysis, performed on the measured stray flux, revealed that the spectral content of the stray flux affected the magnetostrictive behavior of the magnetic dipoles of the terfenol-D transducer, which translated into strain on the fiber gratings.

Evaluation of the Applicability of IFRA for Short Circuit Fault Detection of Stator Windings in Synchronous Machines

A machine is a device to realize the mutual conversion of electric energy and mechanical energy. This work mainly discusses the applicability of impulse frequency response analysis to detect the stator winding short circuit faults. The parameters of excitation waveform and signal sampling were carefully selected. A solid-state nanosecond pulsed power source prototype was developed for resonant excitation of windings. The proposed method was then verified on a 5 kVA, 380V, 2-pole, the salient synchronous machine through artificial fault experiments. Finally, the proposed method was compared with other traditional methods. The method has the advantages of fast measurement, rich high frequencies, high stability, and low cost. Since the principle of impulse frequency response analysis is also applicable to large synchronous machines, this method and device may also be applied to detect large synchronous machines winding short circuit faults. It can be used as an auxiliary and alternative method in the non-destructive fault detection of synchronous machines.

Supervised Classification Of induction Motors faults

Currently, the environmental challenges have been considered as a strategic issue for most industrial companies around the world, threatening their sustainability and profit; This leads to taking the environmental dimensions seriously and preserving natural resources well, since they are a key criterion for sustainable development. In this context, this work calls for innovative solution and new technologies to support the development and integration of environmental considerations through the implementation of an automated fault detection and diagnosis system in induction machines in order to minimize downtime, increase machine utilization rate, get an idea of remaining machine life based on artificial intelligence (AI) and the analysis of collected data. Using the Pattern Recognition methods, this system aims to support decision making in terms of defect classification, through the following process: the collection of relevant data about the stator currents of two induction machines, powered by a converter, one healthy and the other defective, through the CompactRIO device, then the analysis of the data, using programs developed under LabVIEW software, and the extraction of the indicators to form a database. Based on analysis results, several intelligent methods by classification algorithms can organize the acquired data in order to automate the diagnostic process. Ultimately, the set-up of an alert system to prevent rather than cure. The outcomes showed that the integration of predictive maintenance could help achieve an energy cost recovery equal to10% of the total costs of an electric motor system. Hence, the premature detection of faults helps to minimize energy expenditure and achieve overall cost savings, which implies energy optimization.

Machine fault detection methods based on machine learning algorithms: A review.

Preventive identification of mechanical parts failures has always played a crucial role in machine maintenance. Over time, as the processing cycles are repeated, the machinery in the production system is subject to wear with a consequent loss of technical efficiency compared to optimal conditions. These conditions can, in some cases, lead to the breakage of the elements with consequent stoppage of the production process pending the replacement of the element. This situation entails a large loss of turnover on the part of the company. For this reason, it is crucial to be able to predict failures in advance to try to replace the element before its wear can cause a reduction in machine performance. Several systems have recently been developed for the preventive faults detection that use a combination of low-cost sensors and algorithms based on machine learning. In this work the different methodologies for the identification of the most common mechanical failures are examined and the most widely applied algorithms based on machine learning are analyzed: Support Vector Machine (SVM) solutions, Artificial Neural Network (ANN) algorithms, Convolutional Neural Network (CNN) model, Recurrent Neural Network (RNN) applications, and Deep Generative Systems. These topics have been described in detail and the works most appreciated by the scientific community have been reviewed to highlight the strengths in identifying faults and to outline the directions for future challenges.

An Expert System for Rotating Machine Fault Detection Using Vibration Signal Analysis.

Accurate and early detection of machine faults is an important step in the preventive maintenance of industrial enterprises. It is essential to avoid unexpected downtime as well as to ensure the reliability of equipment and safety of humans. In the case of rotating machines, significant information about machine’s health and condition is present in the spectrum of its vibration signal. This work proposes a fault detection system of rotating machines using vibration signal analysis. First, a dataset of 3-dimensional vibration signals is acquired from large induction motors representing healthy and faulty states. The signal conditioning is performed using empirical mode decomposition technique. Next, multi-domain feature extraction is done to obtain various combinations of most discriminant temporal and spectral features from the denoised signals. Finally, the classification step is performed with various kernel settings of multiple classifiers including support vector machines, K-nearest neighbors, decision tree and linear discriminant analysis. The classification results demonstrate that a hybrid combination of time and spectral features, classified using support vector machines with Gaussian kernel achieves the best performance with 98.2% accuracy, 96.6% sensitivity, 100% specificity and 1.8% error rate.

Stray Flux Multi-Sensor for Stator Fault Detection in Synchronous Machines

The aim of this paper is to detect a stator inter-turn short circuit in a synchronous machine through the analysis of the external magnetic field measured by external flux sensors. The paper exploits a methodology previously developed, based on the analysis of the behavior with load variation of sensitive spectral lines issued from two flux sensors positioned at 180° from each other around the machine. Further developments to improve this method were made, in which more than two flux sensors were used to keep a good sensitivity for stator fault detection. The method is based on the Pearson correlation coefficient calculated from sensitive spectral lines at different load operating conditions. Fusion information with belief function is then applied to the correlation coefficients, which enable the detection of an incipient fault in any phase of the machine. The method has the advantage to be fully non-invasive and does not require knowledge of the healthy state.

Fault detection in water pumps based on sound analysis using a deep learning technique

Machine fault detection is designed to automatically detect faults or damage in machines. When a machine operates, it produces vibrations and sound signals that can be analyzed to provide information about the status of the machine. This study proposed a method to detect the faults in a machine based on sound analysis using a deep learning technique. The sound signals generated by the machine were obtained and analyzed under different operating conditions. These signals were first pre-processed to eliminate noise, and then the features were extracted as mel-spectrograms so that the convolutional neural network could automatically learn the appropriate features required for classification. Experiments were conducted on three different water pumps during suction from and discharge to the water tank under normal and abnormal operating conditions. The high accuracies in fault detections in both known and unknown machines indicated that the proposed model performed very well in the detection of machine faults.