Non-invasive Condition Monitoring Research Articles

Comparative study between two artificial intelligence techniques (NN and SVM) applied in fault diagnosis of Wind Turbine

This paper addresses the application of artificial intelligence (AI) techniques for the classification of faults in wind turbine. Wind Energy Conversion Systems have become a focal point in the research of renewable energy sources. Reliability of wind turbine is critical to extract this maximum amount of energy from the wind. Many condition monitoring techniques that are based on steady-state analysis are being applied to wind generators. Bearing faults in this generator causes mechanical vibrations and the variations in the air gap density. The air gap flux density is modulated and currents are generated at different frequencies related to the mechanical vibrations. Characteristic bearing frequencies are associated with the physical construction of bearing and its failure mode. Classical spectral analysis using the Fourier transform was used to detect different bearing failure modes. The magnitudes of the frequencies components formed by the bearing defect are small compared to the rest of the current spectrum. This large difference in magnitude makes difficult the detection and classification of bearing faults. In this paper, we show that the utilization of different approaches of artificial intelligence such as neuronal network (NN) and support vector machines (SVM) can be discriminate different failure modes and gives a good basis for an automatic and non-invasive condition monitoring for wind turbine.

An Experimental Study of the Effects of Cylinder Lubricating Oils on the Vibration Characteristics of a Two-Stroke Low-Speed Marine Diesel Engine

Abstract Two-stroke, low-speed diesel engines are widely used in large ships due to their good performance and fuel economy. However, there have been few studies of the effects of lubricating oils on the vibration of two-stroke, low-speed diesel engines. In this work, the effects of three different lubricating oils on the vibration characteristics of a low-speed engine are investigated, using the frequency domain, time-frequency domain, fast Fourier transform (FFT) and short-time Fourier transform (STFT) methods. The results show that non-invasive condition monitoring of the wear to a cylinder liner in a low-speed marine engine can be successfully achieved based on vibration signals. Both the FFT and STFT methods are capable of capturing information about combustion in the cylinder online in real time, and the STFT method also provides the ability to visualise the results with more comprehensive information. From the online condition monitoring of vibration signals, cylinder lubricants with medium viscosity and medium alkali content are found to have the best wear protection properties. This result is consistent with those of an elemental analysis of cylinder lubrication properties and an analysis of the data measured from a piston lifted from the cylinder after 300 h of engine operation.

Controlled Creation of Contact Cracks in Additive Manufactured Components

Techniques for the controlled seeding and growth of cracks are urgently required for non-destructive testing technique evaluation, particularly for additive manufactured (AM) samples. This paper describes a method that uses a combination of the tensile load and the resonance excitation of notched AM samples, with in situ monitoring of the resonance frequency serving to track the crack dimensions. Mechanical low-cycle fatigue cracks, ranging in length from ~0.3 mm to ~5 mm, were successfully created in five AM samples using this technique. The samples were non-destructively characterized using optical microscopy and Nonlinear Resonance (NLR) testing. The exploitation of resonance enabled the concentration of a significant number of stress cycles on the samples in much shorter timespans than conventional fatigue testing, enabling a high throughput while utilizing compact components. Furthermore, the tracking of the resonance frequency shift throughout the process enabled non-invasive and no-contact real-time condition monitoring.

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.

Condition Monitoring of DC-Link Capacitors Using Time–Frequency Analysis and Machine Learning Classification of Conducted EMI

Condition monitoring techniques for power electronics components are important for reducing maintenance costs and increasing reliability in systems such as aircraft. This article presents a noninvasive condition monitoring system that utilizes time-frequency analysis of conducted electromagnetic interference (EMI) to classify the health of the dc-link capacitor within a three-phase inverter. The approach proposes a combined EMI filter and measurement board which is placed on the dc bus of the inverter. This board filters conducted EMI effectively and enables the inverter to comply with MIL-STD-461 G. It also enables EMI measurements to be collected for condition monitoring applications. The EMI content obtained from this board is analyzed from 15–43 MHz during switching events using a continuous wavelet transform. These characteristic switching images are used to train support vector machine models that are able to classify dc-link health into one of five health stages with accuracy up to 100%.

Reprint of: Few-shot learning for cardiac arrhythmia detection based on electrocardiogram data from wearable devices

Wearable devices have dramatically developed over the past decade as their functions extended from the simple posture analysis to non-invasive condition monitoring for early warning and proactive healthcare, which are especially significant for the dangerous disease such as cardiac arrhythmia. However, it is difficult for the wearable devices to collect plentiful and high-quality training samples so as to meet the fundamental requirements for the learning-based methods. To address this challenge, we propose a meta-transfer based few-shot learning method to handle arrhythmia classification with the ECG signal from the wearable devices. First, the original ECG signals are converted into spectrograms applicable to the 2D-CNN models. Second, we propose the special large-training scheme to pre-train the feature extractor to emphasize the meaningful information for classification, and the feature output dimension is reshaped to reduce the influence of irrelevant and redundant information. Then, the meta-transfer scheme is developed to avoid the training from scratch, which is prone to overfitting without the adequate samples. Finally, we conduct the extensive experiments to assess the performance of our method. The experimental results illustrate that the proposed method outperforms in accuracy than other comparative methods when handling the various few-shot tasks under the same training samples.

Characterization of Fault Signature Due to Combined Air-Gap Eccentricity and Rotor Faults in Induction Motors

An accurate means of non-invasive condition monitoring of the popular industrial drive, three-phase squirrel-cage induction motor, can help to avoid unscheduled maintenance downtime and loss. Faults like air-gap eccentricity can exist even in a newly assembled drive and hence may co-exist with other internal defects. Despite it being a possible situation, the occurrence of simultaneous faults has seldom been studied. Therefore, there is a need for identifying fault signatures of combined fault conditions in a non-invasive manner. This paper presents a detailed model-based study on a three-phase squirrel-cage induction motor with the simultaneous existence of broken rotor-bar and air-gap mixed eccentricity faults using spectral analysis of stator current, instantaneous power, and estimated air-gap torque signals. The modelling of the machine is done using the Multiple Coupled Circuit method and modified to model the presence of the combined fault conditions. A comparative evaluation with various fault conditions and their severity is carried out by spectral analysis, and unique slip-dependent frequency components are identified in the spectra of diagnostic signals. This fault characterization is the most significant contribution of this paper.

Squirrel Cage Induction Motor Defects Diagnosis Using Lissajous Curve of an Auxiliary Winding Voltage

Developing a non-invasive condition monitoring method for a squirrel cage induction machine and its fault diagnosis is the objective of this paper. Several signals can be used to monitor the Induction Machines (IMs) such as power input, stator voltage, and stator current. In this research a new processing signal is developed as an auxiliary winding voltage. The auxiliary winding is considered as a small coil inserting between the stator phases. This approach is based on a Lissajous curve of the auxiliary winding voltage Park component. First, the squirrel cage induction motor is mathematically modeled by considering the real geometry of the rotor structure. After that, the auxiliary winding voltage expression and its Park component are presented. In order to verify the proposed approach effectiveness, the simulation has been carried out in MATLAB for non-defected motor and motor with different load level. The results have confirmed the capability of this method to detect failures.

A Noninvasive Interturn Insulation Condition Monitoring Method Based on the Common-Mode Impedance Spectrum of Inverter-Fed Machines

The winding impedance spectrum method becomes popular in online stator insulation monitoring in recent years. It can detect the insulation ageing condition and give alarm before catastrophic failure happens. However, the theoretical explanation of this method is not sufficient and the relationship between impedance spectrum and insulation ageing is not clear. In addition, the existing methods need additional signal injection equipment, which is intrusive and difficult to be applied in industrial applications. Therefore, this article develops a noninvasive interturn insulation condition monitoring method based on the common-mode impedance spectrum in inverter-fed machines. By building a broadband stator winding insulation model, the relationship between the interturn insulation ageing and the impedance spectrum is studied and the high frequency parallel capacitance is first proposed as an indicator for the interturn ageing. The offline test validate the effectiveness of the proposed method and the online test shows the potential of this method for noninvasive condition monitoring in inverter-fed machines. Future work is planned to further improve the online performance of the proposed method.

A Deep Learninag based Non-invasive and Real-Time Fault Detection System for 3-Phase Induction Motors

Induction motor plays a major role in industry. Despite of its strong structure, induction motors are often prone to faults. There are different types of faults that occurs in the induction motor such as bearing faults, winding faults, etc. Thus motors in major applications require continuous and effective monitoring. In this paper, a stand-alone and non-invasive condition monitoring system that can monitor the condition of 3-phase induction motor using motor current signatures with aid of deep learning (DL) approaches. The proposed system extracts the features using non-invasive current sensors it further samples the features using an analog to digital converter (ADC) and organizes the data acquired from ADC using Raspberry-pi microcomputer. The current data acquired from induction motor is used to train and test the DL models including Multilayer Perceptron (MLP), Long Short-term Memory (LSTM), and One-Dimensional Convolutional Neural Networks (1DCNN). The comparative analysis is demonstrated and the LSTM model as best fault classifier among all with accuracy up to 100%. Finally, the real-time testing of the device showed that the developed system can effectively monitor the conditions of motor using non-invasive current sensors.

Few-shot learning for cardiac arrhythmia detection based on electrocardiogram data from wearable devices

Wearable devices have dramatically developed over the past decade as their functions extended from the simple posture analysis to non-invasive condition monitoring for early warning and proactive healthcare, which are especially significant for the dangerous disease such as cardiac arrhythmia. However, it is difficult for the wearable devices to collect plentiful and high-quality training samples so as to meet the fundamental requirements for the learning-based methods. To address this challenge, we propose a meta-transfer based few-shot learning method to handle arrhythmia classification with the ECG signal from the wearable devices. First, the original ECG signals are converted into spectrograms applicable to the 2D-CNN models. Second, we propose the special large-training scheme to pre-train the feature extractor to emphasize the meaningful information for classification, and the feature output dimension is reshaped to reduce the influence of irrelevant and redundant information. Then, the meta-transfer scheme is developed to avoid the training from scratch, which is prone to overfitting without the adequate samples. Finally, we conduct the extensive experiments to assess the performance of our method. The experimental results illustrate that the proposed method outperforms in accuracy than other comparative methods when handling the various few-shot tasks under the same training samples.

Explainable AI reveals changes in skin microbiome composition linked to phenotypic differences

Alterations in the human microbiome have been observed in a variety of conditions such as asthma, gingivitis, dermatitis and cancer, and much remains to be learned about the links between the microbiome and human health. The fusion of artificial intelligence with rich microbiome datasets can offer an improved understanding of the microbiome’s role in human health. To gain actionable insights it is essential to consider both the predictive power and the transparency of the models by providing explanations for the predictions. We combine the collection of leg skin microbiome samples from two healthy cohorts of women with the application of an explainable artificial intelligence (EAI) approach that provides accurate predictions of phenotypes with explanations. The explanations are expressed in terms of variations in the relative abundance of key microbes that drive the predictions. We predict skin hydration, subject's age, pre/post-menopausal status and smoking status from the leg skin microbiome. The changes in microbial composition linked to skin hydration can accelerate the development of personalized treatments for healthy skin, while those associated with age may offer insights into the skin aging process. The leg microbiome signatures associated with smoking and menopausal status are consistent with previous findings from oral/respiratory tract microbiomes and vaginal/gut microbiomes respectively. This suggests that easily accessible microbiome samples could be used to investigate health-related phenotypes, offering potential for non-invasive diagnosis and condition monitoring. Our EAI approach sets the stage for new work focused on understanding the complex relationships between microbial communities and phenotypes. Our approach can be applied to predict any condition from microbiome samples and has the potential to accelerate the development of microbiome-based personalized therapeutics and non-invasive diagnostics.

A Digital Twin Based Estimation Method for Health Indicators of DC–DC Converters

This article proposes a health indicator estimation method based on the digital-twin concept aiming for condition monitoring of power electronic converters. The method is noninvasive, without additional hardware circuits, and calibration requirements. An application for a buck dc-dc converter is demonstrated with theoretical analyses, practical considerations, and experimental verifications. The digital replica of an experimental prototype is established, which includes the power stage, sampling circuit, and close-loop controller. Particle swarm optimization algorithm is applied to estimate the unknown circuit parameters of interest based on the incoming data from both the digital twin and the physical prototype. Cluster-data of the estimated health indicators under different testing conditions of the buck converter is analyzed and used for observing the degradation trends of key components, such as capacitor and MOSFET. The outcomes of this article serve as a key step for achieving noninvasive, cost-effective, and robust condition monitoring for power electronic converters.

A deep learning approach to condition monitoring of cantilever beams via time-frequency extended signatures

We introduce with this work a deep learning approach for non-invasive condition monitoring of cantilever beams. The deep learning classifier is used to recognize a damaged or undamaged beam via time-frequency extended signatures. These signatures are the distributions over several measurements of the natural frequencies extracted from the refined time-frequency adaptive spectrum of vibrating beams. The test results showed that we are able to cancel ambient effects like the temperature and to obtain a high accuracy of the results which for the considered cases reach 100%.

Non-invasive condition monitoring for boost converter based on crow search algorithm

Non-invasive condition monitoring for boost converter based on crow search algorithm

A Novel Non-intrusive Method to Diagnose Bearings Surface Roughness Faults in Induction Motors

Noninvasive condition monitoring methods are economical and easy to implement. An extensive research has been conducted in the past to diagnose bearing localized faults through noninvasive condition monitoring techniques. However, due to unavailability of the harmonic frequency model for bearing surface roughness faults, noninvasive techniques could not be used to diagnose such type of faults. This paper aims to mathematically derive a harmonic frequency model for bearing surface roughness faults using bearing geometry. The derived harmonic frequency model has been verified experimentally using noninvasive instantaneous power analysis technique. Laboratory tests have been performed on the inner and outer race surface roughness faults under no-load and full-load conditions to validate the approach.

Analysis of distributed faults in inner and outer race of bearing via Park vector analysis method

The Park’s transformation technique for diagnosing and statistically analyzing a variety of bearing faults is introduced in this paper. The currently used stator current analysis and instantaneous power analysis methods are not capable of diagnosing bearing distributed faults, because the defect frequency model is not available for this kind of faults. Notably, this paper has been aimed at developing a system for the non-invasive condition monitoring of bearing distributed defects on the basis of the Park vector analysis. It is also aimed at statistically evaluating the ability of this developed system to not only analyze but also segregate the localized and distributed faults in the bearings. The theoretical as well as experimental work that has been carried out demonstrates that the proposed technique can not only diagnose both types of bearing faults, but also classify them. The effectiveness of the proposed technique has been confirmed by the results obtained from real hardware implementation.

An Online Non-Invasive Condition Monitoring Method for Stepping Motor CRDM in HTGR

Control Rod Drive Mechanism (CRDM) based on stepping motor is one of the components applied in High Temperature Gas Coold Reactor (HTGR) to control the reactivity as well as to maintain the safety of reactor. The stepping motor requires a unique condition monitoring to avoid any failures especially due to the specific environments of CRDM in HTGR such as the allowable of high temperature, high radiation and the location of stepper motor inside a pressure shell. This research aims to demonstrate an online non-invasive condition monitoring method without direct access to the CRDM of HTGR based on voltage and stator current measurements. A simple stepping motor CRDM simulator is employed. The online condition monitoring is carried out by direct pattern matching of the output signals of logic generator block and the output signals of motor driver. The online method utilizes signature patterns of voltage and stator current signals of the healthy motor as a baseline for healthy motor. In addition, the method is applied to detect high-resistance problem on the connector between the motor driver block and the stepper motor to show the effectiveness and the applicability of this method. The online condition monitoring system demonstrates a capability to identify a minimum detectable simulated high-resistance for about 2.9% which decreases the measured stator current and motor’s torque for around 5.1% and 3.3%, respectively. The paper also points out signatures of healthy motor, including mutual inductions of the motor’s winding in voltage and current measurement which can be used as the fault symptom indicators for online monitoring purposes. Received: 19 November 2014; Revised: 13 April 2016; Accepted: 22 June 2016

Planetary Gearbox Torsional Vibration Effects on Wound-Rotor Induction Generator Electrical Signatures

In this paper, torsional vibration effects of a planetary gearbox on electrical signatures of a wound-rotor induction generator have been studied in a representative complex electromechanical system through both numerical simulations and experiments. A universal lumped-parameter model has been developed with consideration of impacts of main drive-train mechanical components. This model has been adapted to a reduced-scale wind turbine simulator based on a 5.5-kW three-phase wound-rotor induction generator used for experimental validation at different load conditions. For this system, it has been demonstrated that most of mechanical characteristic frequencies of the planetary gearbox are difficult to be identified in the stator current spectrum. Nevertheless, the proposed approach has been used as an initial stage to assess the effectiveness of noninvasive condition monitoring of a planetary gearbox based on electrical signatures analysis.

A PLL-based resampling technique for vibration analysis in variable-speed wind turbines with PMSG: A bearing fault case

Condition monitoring in permanent magnet synchronous machines has gained interest due to the increasing use in applications such as electric traction and power generation. Particularly in wind power generation, non-invasive condition monitoring techniques are of great importance. Usually, in such applications the access to the generator is complex and costly, while unexpected breakdowns results in high repair costs. This paper presents a technique which allows using vibration analysis for bearing fault detection in permanent magnet synchronous generators used in wind turbines. Given that in wind power applications the generator rotational speed may vary during normal operation, it is necessary to use special sampling techniques to apply spectral analysis of mechanical vibrations. In this work, a resampling technique based on order tracking without measuring the rotor position is proposed. To synchronize sampling with rotor position, an estimation of the rotor position obtained from the angle of the voltage vector is proposed. This angle is obtained from a phase-locked loop synchronized with the generator voltages. The proposed strategy is validated by laboratory experimental results obtained from a permanent magnet synchronous generator. Results with single point defects in the outer race of a bearing under variable speed and load conditions are presented.