No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Reviewing Fault Diagnosis Methods in Electric Drives: Power Subsystem and Electrical Machine
Modern industrial and commercial electrical drives are typically designed to control motors of different types, operating under scalar, field-oriented or direct torque control. These drives execute various algorithm such as sensorless control, temperature monitoring, etc., which require the knowledge of motor drive parameters. In order to do this, the inverters allow data to be input manually or execute self-commissioning routines before running the motor. The parameters necessary for proper operation include motor phase resistance and inverter voltage drop, which are especially important in low-speed range. This paper presents an offline technique for estimation of mentioned parameters, which can be obtained with enough precision for the overwhelming majority of applications. The proposed algorithm consequently injects DC-current of several levels into the stator of a motor, measuring the corresponding voltages. Each pair of current and voltage is a point in the voltage current characteristic of motor drive, thus set of these points maybe approximated with a first order polynomial using least squares method. The parameters of polynomial are desired inverter voltage drop and phase resistance. The experimental section analyzes estimation errors and their dependence on the number of injected levels, their values and filtering capability of measuring algorithm. After that, the authors give suggestions on algorithm parameters selection, depending on the demanded precision. Finally, the authors demonstrate a mass-producing dishwasher motor drive, which adopted this technique.
The paper develops accurate analytical subdomain models for predicting the magnetic and armature reaction fields in fault-tolerant flux-switching permanent-magnet machines. The entire region is divided into five subdomains, followed by rotor slots, air-gap, stator slots, PM, and external air-gap imported to account for flux leakage. The coil turns and the remanence of magnets are adjusted by keeping the magnetic and electrical loading on the motor constant. The distance between the centers of two adjacent stator slots varies due to the introduction of fault-tolerant teeth. According to the variable separation method, the general solution expression of each region can be determined by solving the partial differential systems of equations. The magnetic field distributions of subdomains are obtained by applying the continuity conditions between adjacent regions. Some analytical field expressions are represented as new forms under armature reaction field condition compared to those under no-load condition. Based on the developed analytical models, the flux density distribution and the electromagnetic performance can be calculated under no-load or armature reaction field condition separately. The finite element analysis is carried out to verify the validity of the proposed analytical model.
Two-three (2/3)-level dual-active-bridge (DAB) converter is a promising DC-DC converter for medium and high voltage applications. That is due to its advantages like high power density, galvanic isolation, and capability of withstanding higher voltage ratings compared to the two-level DAB converters. However, the open-circuit fault (OCF) is critical for the 2/3-level DAB converters, resulting in various negative effects, e.g., DC bias, overshoot current, and capacitor voltage imbalance. To address these issues, it is necessary to develop fault diagnosis and fault-tolerant control strategies. This paper thus proposes a method to identify the faulty switch based on the dynamic characteristics when the OCF occurs. The midpoint voltages of the neutral-point-clamped (NPC) bridge arms are employed as the fault diagnosis signals, where the faulty switch can be identified accurately based on the mean values and duty cycles of the midpoint voltages. Subsequently, a fault-tolerant control scheme based on a complementary-switch-blocking (CSB) method is proposed. In this scheme, when the OCF occurs on one switch, the gate-driving signal of its complementary switch is blocked, and the OCF negative effects, e.g., DC bias, overshoot current, and unbalanced capacitor voltages, can be alleviated significantly. Furthermore, the power transfer capability of the 2/3-level DAB converters is enhanced with the proposed scheme compared to the traditional bypass-arm (BPA) method. Finally, experimental tests are carried out to verify the theoretical analysis.
Electrical machines are prone to faults and failures and demand incessant monitoring for their confined and reliable operations. A failure in electrical machines may cause unexpected interruptions and require a timely inspection of abnormal conditions in rotating electric machines. This article aims to summarize an up-to-date overview of all types of bearing faults diagnostic techniques by subdividing them into different categories. Different fault detection and diagnosis (FDD) techniques are discussed briefly for prognosis of numerous bearing faults that frequently occur in rotating machines. Conventional approaches, statistical approaches, and artificial intelligence-based architectures such as machine learning and deep learning are discussed summarily for the diagnosis of bearing faults that frequently arise in revolving electrical machines. The most advanced trends for diagnoses of frequent bearing faults based on intelligence and novel applications are reviewed. Future research directions that are helpful to enhance the performance of conventional, statistical, and artificial intelligence (machine learning, deep learning) and novel approaches are well addressed and provide hints for future work.
Bearings, as the key mechanical components of rotary machinery, are widely used in modern aerospace equipment, such as helicopters and aero-engines. Intelligent fault diagnosis, as the main function of prognostic health management systems, plays a critical role in maintaining equipment safety in aerospace applications. Recently, data-driven intelligent diagnosis approaches have achieved great success due to the availability of large-scale, high-quality, and complete labeled data. However, in a real application, labeled data is often scarce because it requires manual labeling, which is time-consuming and labor-intensive. Meanwhile, health monitoring data are usually scattered in different regions or equipment in the form of data islands. Traditional fault diagnosis techniques fail to gather enough data for model training due to data security, economic conflict, relative laws, and other reasons. Therefore, it is a challenge to effectively combine the data advantages of different equipment to develop an intelligent diagnosis model with better performance. To address this issue, a novel clustering federated learning (CFL) method with a self-attention mechanism is proposed for bearing fault diagnosis. Firstly, a deep neural network with a self-attention mechanism is developed in a convolutional pipe for feature extraction, which can capture local and global information from raw input. Then, the CFL is further constructed to gather the data from different equipment with similar data distribution in an unsupervised manner. Finally, the CFL-based diagnosis model can be well trained by fully utilizing the distributed data, while ensuring data privacy safety. Experiments are carried out with three different bearing datasets in aerospace applications. The effectiveness and the superiority of the proposed method have been validated compared with other popular fault diagnosis schemes.
The issues of monitoring and fault diagnosis of drives with permanent magnet synchronous motors (PMSMs) are currently very relevant because of the increasing use of these drives in safety-critical devices. Every year, more and more articles are published on this subject. Therefore, the objective of this article is to update the overview of diagnostic methods and techniques for PMSM drives. Each of the main chapters of the article focuses on a specific element of the drive system (motor, power converter, measuring sensors), with particular emphasis on the components of the motor (stator windings, magnets, bearings, and rotor). The main sections on PMSM fault diagnosis are divided according to the type of methods used to obtain the symptoms of the damage. In addition, a review of methods that use the analysis of signals of the control structure for the diagnosis of damage to a vector-controlled motor is presented, as well as the latest achievements of researchers in the field of shallow and deep neural networks for the detection and classification of failures of PMSM drives. Based on the analyses presented in the literature, some development trends and challenges related to the development of diagnostics and fault-tolerant control of PMSM drives are discussed in the conclusion part.
As technology advances, the utilization of lighting systems based on light-emitting diode (LED) technology is becoming increasingly essential, given its benefits in terms of efficiency, reliability, and lifespan. Unfortunately, the power electronic components required to drive LEDs are unable to compete with LED devices in terms of lifetime. Aluminum electrolytic capacitor (AEC) failures represent the root cause of power electronic equipment breakdown, mainly through both aging and temperature effects. This highlights the importance of designing robust power converter architectures and control methods that allow the evaluation of the condition of electrolytic capacitors while maintaining the performance of converter controllers, even in the presence of capacitor failure. On this basis, this work proposes a novel condition-monitoring system for the diagnosis of capacitor faults on a fault-tolerant LED driver, which is able to deal with the specific architecture and low ratings of the most recent LED lighting systems. The fault-detection task applies the short time least square Prony’s (STLSP) approach to perform an online estimation of the ESR and C parameters, allowing the continuous evaluation of the electrolytic capacitor’s condition and, as a result, the prevention of total system failure. With regard to capacitor failure, the experimental results suggest that the condition-monitoring task is extremely effective, even when considering a limited number of data samples.
Electric powertrain is constituted by electric machine transmission unit, inverter and battery packs, etc., is a highly-integrated system. Its reliability and safety are not only related to industrial costs, but more importantly to the safety of human life. This review is the first contribution to comprehensively summarize both the feature engineering methods and artificial intelligence (AI) algorithms (including machine learning, neural networks and deep learning) in electric powertrain condition monitoring and fault diagnosis approaches. Specifically, this paper systematically divides the AI-supported method into two main steps: feature engineering and AI approach. On the one hand, it introduces the data and feature processing in AI-supported methods, and on the other hand it summarizes input signals, feature methods and AI algorithms included in the AI method in cases. Therefore, firstly this review is to guide how to choose the appropriate feature engineering method in further research. Secondly, the up-to-date AI algorithms adopted for powertrain health monitoring are presented in detail. Finally, such current approaches are discussed and future trends are proposed.
Electrical machines are to face different challenging factors during operation, such as high unexpected or excessive loads, unusual properties of the working environment, or intense fluctuations in rotation speed. Therefore, maintenance questions and predicting the accuracy of an equipment’s condition have great importance. This study is based on the theory of vibration reliability. This article introduces the most common faults of bearings in electrical machines and discusses their diagnostic possibilities. Experimental setup, as well as studied bearing failures, are described. The accuracy of conducted experiments is introduced.
In engineering, the fault data unevenly distribute and difficultly share, which causes that the existing fault diagnosis methods cannot recognize the newly added fault types. An intelligent diagnosis method for machine fault is proposed based on federated learning. Firstly, the local fault diagnosis models diagnosing the existing fault data and the newly added fault data are established by deep convolutional neural network. Then, the weight parameters of local models are fused into global model parameters by federated learning. Finally, the global model parameters are transmitted to each local model. Therefore, each local model update into a global shared model which can recognize the newly added fault types. The proposed method is verified by bearing data. Compared with the traditional model, which can only diagnose the existing fault data but cannot recognize newly added fault types, the federated fault diagnosis model fusing weight parameters can diagnose newly added faults without exchanging the data, and the accuracy is 100%. The proposed method provides an effective method to solve the poor sharing of fault data and poor generalization of fault diagnosis model for mechanical equipment.
Bearing fault diagnosis can be used to accurately and automatically identify the type and severity of faults. Federation learning can perform learning without transferring local data among multiple local nodes with the same data features. The traditional federated learning algorithm is difficult to identify high-quality local models among class-unbalanced local nodes, which leads to the poor quality of the training model and the slow training speed. To improve the quality and training speed of the model, this paper proposes the FA-FedAvg algorithm for fault diagnosis based on the traditional federated learning algorithm. Specifically, the weighting strategy of the model is optimized, which is conducive to increasing the weight of high-quality local models, thereby improving the quality of training models. Then, a model aggregation strategy based on precision difference is proposed to reduce the number of iterations and accelerate the convergence of the training model. Finally, the proposed algorithm is compared with FedAvg and FedProx algorithms under different data distribution conditions. The experimental results show that, compared with the comparison algorithm, the number of model training iterations of the FA-FedAvg algorithm is reduced by 52.5% on average, and the fault classification accuracy has an average increment of about 8.6%. Moreover, the FA-FedAvg fault diagnosis method is robust under different classes and data volumes.
This paper discusses open phase faults, which can happen in a system grid-converter-motor drive, and measures to properly handle this type of fault. The authors of this paper provide classification of algorithms used for the detection of the loss of phase and explain the distinctive features of each group of methods. They review existing algorithms, which are based on the analysis of the current signals, discuss their pros and cons and suggest possible areas of usage for each group of methods. The authors also propose one novel method for detection of open phase, which was developed for low-cost systems with low resolution analog-to digital converters (ADC). This paper mainly considers methods for three-phase motors as the most popular machines, however some algorithms can be used in multiphase drives. The authors of the paper also share their more than 20 years’ experience combined, in this area, which was obtained by developing industrial and commercial drives, and focus on the requirements of the IEC/UL 60730 safety standard, where the phase-loss detection algorithm is one of the essential parts of control system.
A review of the fault diagnostic techniques based on machine learning is presented in this paper. As the world is moving towards industry 4.0 standards, the problems of limited computational power and available memory are decreasing day by day. A significant amount of data with a variety of faulty conditions of electrical machines working under different environments can be handled remotely using cloud computation. Moreover, the mathematical models of electrical machines can be utilized for the training of AI algorithms. This is true because the collection of big data is a challenging task for the industry and laboratory because of related limited resources. In this paper, some promising machine learning-based diagnostic techniques are presented in the perspective of their attributes.
Switched reluctance motor drives are a common technology used in traction motor drives. Herein, an online method is proposed for the fault diagnosis of power transistors in the popular asymmetric half‐bridge power converter of the SRM drive. Based on the rearrangement of three current sensors, each phase current was first calculated by solving equations associated with their detected values and the drive signals of the transistors. The faults in the transistors were then preliminarily detected by monitoring the error between the calculated sum of currents and the sum of actual phase currents detected by a current sensor. Once a fault was identified, the actual states of all transistors of the power converter were solved for inversely using a mathematical model and the necessary rule for a trade‐off. Then, the faulty transistors and the fault types were identified by comparing the actual states with the drive signals. Compared with prevalent methods, the diagnostic range of the proposed scheme was wider, and its control modes and the number of motor phases were not limited. A higher accuracy than currently available methods was its prominent advantage. The effectiveness of the proposed solution was also validated on a three‐phase 6/4 SRM drive.
Multiple sensor data fusion is necessary for effective condition monitoring as the electric machines operate in a wide range of diverse operations. This study investigates sensor acquired vibration and current signals to establish a reliable multi-fault diagnosis framework of a brushless DC (BLDC) motor. Faults in stator and rotor were created deliberately by shorting two adjacent windings and creating a hole on the surface, respectively. The threshold for different health states was obtained by the third harmonic analysis of motor current. Later, the key features from sensor acquired current and vibration signals are selected based on monotonicity and reduced using the principal component analysis (PCA). For future predictions, an artificial neural network (ANN) is used to classify different fault features and its performance is evaluated using several metrics. Analysis of motor current harmonics and impulsive vibration response at the same time provides a thorough health estimation of BLDC motor in the presence of both electrical and mechanical faults. Multiple sensor information is fused to obtain a better understanding of the fault characteristics and mitigate the randomness of fault diagnosis. The proposed model was able to detect and classify multiple fault features with higher accuracy compared to other similar methods.
Applying the fault diagnosis techniques to power electronic equipments is beneficial to improve the stability and reliability of renewable energy system, because power electronic equipments are a key component of renewable energy system that largely defines their performance. This paper presents a novel intelligent fault diagnosis method for a three-phase power-electronics energy conversion system based on knowledge-based and data-driven methods. Firstly, the three-phase AC currents of the power-electronics energy conversion system are collected and used to analyze. Secondly, the feature transformation, a knowledge-based method, is utilized to preprocess the fault data. After feature transformation, the slopes of current trajectories (transformed features) are not affected by different loads. And then random forests algorithms (RFs), a data-driven method, are adopted to train the fault diagnosis classifier with the processed fault data. Finally, the proposed method is implemented online on an actual three-phase PWM rectifier platform. The results show that the proposed fault diagnosis method can successfully detect and locate the open-circuit faults in IGBTs of the three-phase PWM rectifier. In addition, the proposed method is suitable for most of three-phase power-electronics energy conversion systems.
This paper proposes a fault-detection method for open-switch failures in hybrid active neutral-point-clamped (HANPC) rectifiers. The basic HANPC topology comprises two SiC-based metal-oxide-semiconductor field-effect transistors (MOSFETs) and four Si insulated-gate bipolar transistors (IGBTs). A three-phase rectifier system using the HANPC topology can produce higher efficiency and lower current harmonics. An open-switch fault in a HANPC rectifier can be a MOSFET or IGBT fault. In this work, faulty cases of six different switches are analyzed based on the current distortion in the stationary reference frame. Open faults in MOSFET switches cause immediate and remarkable current distortions, whereas, open faults in IGBT switches are difficult to detect using conventional methods. To detect an IGBT fault, the proposed detection method utilizes some of the reactive power in a certain period to make an important difference, using the direct-quadrant (dq)-axis current information derived from the three-phase current. Thus, the proposed detection method is based on three-phase current measurements and does not use additional hardware. By analyzing the individual characteristics of each switch failure, the failed switch can be located exactly. The effectiveness and feasibility of the proposed fault-detection method are verified through PSIM simulations and experimental results.
In order to not change the space vector pulse width modulation (SVPWM) control strategy during one phase fault, the five-phase six-bridge arm SVPWM fault tolerant control method for fifteen-phase permanent magnet synchronous motor (PMSM) is proposed in this paper, and the thermal stress of fifteen-phase PMSM under different fault-tolerant operations is analyzed. Firstly, the control model of the fifteen-phase PMSM based on three dq axes is established, the generation mode of the SVPWM is analyzed, and the speed and current loop PI regulators of the control system are designed. Secondly, the fault-tolerant control principle of the five-phase six-bridge arm is analyzed and compared with the hysteresis control strategy of equal amplitude and minimum stator loss. Thirdly, the 3D model of the fifteen-phase PMSM is established, the steady-state temperature and the transient temperature rise considering operating conditions under different fault tolerant operations are analyzed, and corresponding temperature rise results of the stator armature windings are compared separately. Finally, the experimental platform is established, the phase current waveforms tested under load conditions confirm the theoretical analysis of five-phase six-bridge arms and hysteresis control, and the test results of steady-state and transient temperature rise confirm the correctness of the simulation prediction.
Despite the emerging multi-phase and multi-level converters, two-level insulated gate bipolar transistor-based power converters are still widely used in industrial applications in nowadays. Thus, its reliability is of significant importance for ensuring industrial safety. In this paper, a review of the possibilities to ensure the reliable operation of these power converters is presented. The possible approaches are categorized into two groups: condition monitoring and fault-tolerant control. The former approach performs the monitoring methods of power converters, which enables the identification of device degradation to be realized. Accordingly, the mechanisms, indicators, and measuring methods of degradation are demonstrated in this paper to assist the design of condition-based maintenance. In contrast, the latter approach is a post-fault one, where power converters remain the operation by activating the fault-tolerant units after faults are identified. For this approach, fault detection, fault isolation, and tolerant strategies are essential. Finally, the performance and cost-effectiveness of the two categories are discussed in this paper.
Nowadays, power electronic technology is widely affecting people’s daily work and life. However, there are still many problems in the current power supply research. When the fault information of power transformer is not complete or there is some ambiguity or even the information is lost, it will largely lead to the conclusion and correct conclusion of fault diagnosis. In this case, the fuzzy theory is applied to the fault diagnosis of shunt capacitor, and the fuzzy fault diagnosis system of shunt capacitor is studied. At the same time, a map-based fault diagnosis system is proposed. In this paper, the cloud computing technology is introduced into the deep learning and compared with SVM and DBN algorithm. The research results of this paper show that the accuracy of fuzzy diagnosis results is 94%, 84%, 90%, 80%, 83%, and 70%, respectively, which shows that the model diagnosis reliability is relatively high. Among the three algorithms, MR-DBN overall detection rate is higher and the time-consuming is lower than the other two methods. The diagnostic accuracy and misjudgment rate of DBN are as follows: 96.33% and 3.90%. The diagnosis accuracy and misjudgment rate of SVM are as follows: 96.40% and 3.83%. The diagnostic accuracy and misjudgment rate of MR-DBN are, respectively, 99.52% and 0.57%. Compared with the other two methods, MR-DBN has the highest diagnostic accuracy and the lowest error rate, which to a large extent indicates that MR-DBN algorithm has higher diagnostic accuracy and has greater advantages and reliability in power supply diagnosis and identification. It not only improves the accuracy of power capacitor fault diagnosis and identification but also provides a new method for the application of power capacitor fault research and development.
The traditional deep learning-based bearing fault diagnosis approaches assume that the training and test data follow the same distribution. This assumption, however, is not always true for the bearing data collected in practical scenarios, leading to significant decline to fault diagnosis performance. In order to satisfy this assumption, the transfer learning concept is introduced in deep learning by transferring the knowledge learned from other data or models. Due to the excellent capability of feature learning and domain transfer, deep transfer learning methods have gained widespread attention in bearing fault diagnosis in recent years. This article presents a comprehensive review of the development of deep transfer learning-based bearing fault diagnosis approaches since 2016. In this review, a novel taxonomy of deep transfer learning-based bearing fault diagnosis methods is proposed from the perspective of target domain data properties divided by labels, machines, and faults. By covering the whole life cycle of deep transfer learning-based fault diagnosis and discussing the research challenges and opportunities, this review provides a systematic guideline for researchers and practitioners to efficiently identify suitable deep transfer learning models based on the actual problems encountered in bearing fault diagnosis.
Motor current signature analysis (MCSA) techniques measure slip-related current signals in the supply lines of a squirrel-cage induction machine (SCIM) for slip estimation, fault detection, and motor diagnostics. In a traditional setup for MCSA, current sensors directly measure the machine currents and are usually installed physically close to the machine of interest, e.g., current transducers clamped about the machine stator leads. For grid-connected SCIMs, the slip-related current signals propagate through the power system. Thus, a single aggregate power monitor can potentially perform MCSA for a collection of SCIMs powered by a common electrical service, referred to in this paper as nonintrusive MCSA. Current division and system impedances (source impedance, load impedances, and machine parameters) determine the applicability of nonintrusive MCSA. This paper demonstrates measurement hardware, modeling, and experimental results for nonintrusive MCSA. Analysis and results investigate the current-source model of a grid-connected SCIM and its interaction with the system impedances. Applied to a distribution system with multiple machines, the techniques inform when nonintrusive MCSA is possible for collections of SCIMs and equivalent to traditional MCSA.
Power electronic converters have become an essential element in a wide variety of applications, such as electric drives, distributed power generation, electric vehicles, among others. Failures in electronic converters lead to operational and safety difficulties, therefore, detection and diagnosis contribute to the reliability of the system, which is part. The aim of this article is to review fault-detection and diagnosis methods in power converters based on electrical measurements. For this, a systematic review is used to establish the main faults, methods, applications, and challenges in power electronic converters. In fact, methods consulted are focused on the capacitor at 20.6%, the switching transistor at 60.3%, and other components at 19.1%.
Traction motor bearings are critical research objects of motor fault diagnosis due to their high failure rate. Nowadays, numerous frequency-domain analysis methods have been proposed for rolling bearing fault diagnosis. However, most methods have rigorous requirements on the applicable conditions, and may fail to diagnose when working conditions change, which significantly limits the broad application of the method in practical engineering. Therefore, this paper presents an adaptative fault diagnosis and decision-making method (AFDDM) based on multi-spectrum evaluation and fusion. The method can give appropriate weights to each spectrum in different conditions to ensure the stability of fused spectrum and convert the spectrums into more intuitive fault probability to give maintenance advice. And the AFDDM method achieves the purpose of evaluating the spectrum by extracting the eigenvalue vector and utilizing the Local Tangent Space Alignment (LTSA) to reduce dimension. Then, a decision-making model based on multi-spectrum is proposed to divert the spectrums into fault probability. By this method, the evaluation fusion of spectrums with different effect and decision making are realized. Real bearing fault signals from different data sets are selected to verify the effectiveness of the proposed method and show that the proposed method has better engineering applicability and robustness.
The peripheral magnetic field of the motor may cause magnetic field interference to the surrounding sensors, which would reduce the accuracy of sensors and affect the safety and concealment of vessel. This article explored the environmental magnetic field distribution law of a permanent magnet (PM) motor with two-pole parallel magnetization. A no-cogging magnetic field model was obtained by solving Maxwell equations, and the relationship between the air gap flux density and the environment flux density was built considering the motor cogging effect. In addition, a mathematical model of connections between the back electromotive force (EMF) and the environmental flux density was built, and the correctness of the model was verified by COMSOL. It is concluded that the circumferential and radial components of the environmental magnetic field are distributed sinusoidally, and the magnetic flux density modulus remains constant on the equidistant circumferential line, but attenuates squarely with the distance from the motor. It provides ideas for magnetic field analysis in the peripheral environment of other types of motors, sensors, and other anti-motor electromagnetic interference, magnetic field compensation for geomagnetic measurement, and motor fault diagnosis.
The modular multilevel converter (MMC) is one of the most commonly used power electronic converters in medium and high-power applications. However, reliability is one of the most critical challenges because of the many power switching devices applied in the MMC. Also, many voltage sensors in the converter topology have increased cost and hardware complexity. In this paper, a method for detecting IGBTs open-circuit fault is proposed. In the proposed method, each converter arm is divided into sets. Each set includes two sub-modules and a voltage sensor. These voltage sensors are responsible for monitoring their set. By comparing the output voltage of each set, in normal operation mode and after the open-circuit fault, the fault is detected with great speed and accuracy. Finally, the performance of the proposed method in various scenarios is examined by the simulation results in MATLAB/Simulink.
Most of the mechanical fault diagnosis methods of induction motors (IMs) are based on vibration signals. However, vibration sensors are expensive and require direct contact with IMs. This paper proposes a data-driven mechanical fault diagnosis method for IMs using stator current signals, which is more convenient and low cost since no additional sensors are required. Aiming at the weak representation of mechanical faults in current signals, an intelligent noise elimination method based on noise reconstruction model is proposed to improve the signal-to-noise ratio. Through the automatic feature extraction and classification of the residual current envelope spectrum, high diagnosis accuracy can be obtained even if some differences are existed among samples. The effectiveness of the proposed method is verified by high accuracy diagnosis results on two experimental platforms. The results show that the average diagnostic accuracy of the proposed algorithm for one bearing fault and two eccentricity faults can reach 96%. Even if the fault type and the fault degree are distinguished at the same time, the accuracy of 90% can be achieved for six kinds of bearing faults.
The open-circuit fault detection and localization (FDL) technique can improve the reliability of the modular multilevel converter (MMC). However, the conventional software-based FDL methods usually have a heavy computation burden or a limited localization speed. This article proposes a simplified and fast software-based FDL approach for the grid-connected MMC. First, the errors between the measured state variables (the output current and the circulating current) and their estimated values are calculated. By comparing these errors with their threshold values, the switch fault can not only be detected but also be localized to the specific arm. Then, the capacitor voltages in this faulty arm are collected, and the submodule (SM) with the highest capacitor voltage is selected. To confirm the switch fault in this SM, a modified
Pauta
criterion is presented to check the abnormal voltage data. As a result, the computation burden of the proposed software-based FDL approach is significantly reduced, and the faulty SM can be localized in a short period. Simulation and experimental results verify that the proposed approach can effectively detect and localize different open-circuit faults, and it is immune to the step of power references.
Early fault detection in power electronic systems (PESs) to maintain reliability is one of the most important issues that has been significantly addressed in recent years. In this paper, after reviewing various literature based on fault detection in PESs, data mining-based techniques including artificial neural network, machine learning, and deep learning algorithms are introduced. Then, the fault detection routine in PESs is expressed by introducing signal measurement sensors and how to extract the feature from it. Finally, based on studies, the performance of various data mining methods in detecting PESs faults is evaluated. The results of evaluations show that the deep learning-based techniques given the ability of feature extraction from measured signals are significantly more effective than other methods and as an ideal tool for future applications in power electronics industry are introduced. Index Terms-Power electronic systems, reliability, fault detection, fault tolerant, artificial neural network, machine learning, deep learning.
The motor drive system plays a significant role in the safety of electric vehicles as a bridge for power transmission. Meanwhile, to enhance the efficiency and stability of the drive system, more and more studies based on AI technology are devoted to the fault detection and diagnosis of the motor drive system. This paper reviews the application of AI techniques in motor fault detection and diagnosis in recent years. AI-based FDD is divided into two main steps: feature extraction and fault classification. The application of different signal processing methods in feature extraction is discussed. In particular, the application of traditional machine learning and deep learning algorithms for fault classification is presented in detail. In addition, the characteristics of all techniques reviewed are summarized. Finally, the latest developments, research gaps and future challenges in fault monitoring and diagnosis of motor faults are discussed.
Modular multilevel converters (MMCs) play an essential role in power conversion for medium and high voltage industrial applications. These types of converters use many sensors for the process of measuring and sorting the voltage of their capacitors, but this increases cost and hardware complexity, as well as decreases reliability. One of the most widely used methods for voltage sorting of capacitors in MMCs is based on current measuring and detecting the current sign of each arm. This paper uses the capacitor voltages estimation scheme by the Kalman filter (KF) algorithm at each sampling time. Moreover, with the difference between the total voltage of the capacitors in each arm during two consecutive processes, their sorting is done without the current sensor. Therefore, by the proposed scheme, current sensors are entirely removed from the arms, and the MMC control process is implemented by only two voltage sensors in a phase branch. The simulation results in the MATLAB Simulink environment confirm the accuracy of the proposed method.
A large amount of labeled data is important to enhance the performance of deep learning based methods in the area of fault diagnosis. Because It is difficult to obtain high-quality samples in real industrial applications, federated learning (FL) is an effective framework for solving the problem of sparse samples by using the distributed data.Its global model is updated by the local client without sharing data at each round. Considering computing resources and communication loss of multiple clients, an efficient method based on Stacked Sparse Autoencoders (SSAEs) and Siamese networks is proposed to detect interturn short circuit (ITSC) faults in permanent magnet synchronous motors (PMSMs). To achieve an accurate ITSC fault detection, a SSAE was employed to extract sparse features in a limited number of samples, and Siamese networks were used to determine the similarity between the given samples. The problem of fault diagnosis is transformed into a classification problem under few-shot learning. Furthermore, the proposed method is trained successfully in the frameworks of centralized learning and decentralized structure. The experimental results indicate that the proposed method achieved high fault diagnosis accuracy. Moreover, it is suitable for deployment in smart manufacturing systems.
The use of variable speed drives made essential the presence of power electronic in electrical systems. The AC to DC and DC to AC conversions thanks to rectifiers and inverters cause difficulties in the ground fault detection. In this paper, a new method to detect ground faults in variable speed drives is presented. It is based on the analysis of the terminal voltage signal of a grounding resistor connected between the neutral at the secondary of the main power transformer and ground. Attending to the waveform, the ground fault can be detected on the AC grid side, DC positive or negative terminals or on the AC inverter side. To verify the method, numerous simulations and experimental tests in a 140 kW power converter were carried out achieving satisfactory results.
Rotating machines are widely used in Industries, Manufacturing and Oil & Gas plants as a critical component for process availability. The inadvertent failure of these rotating machines causes significant process downtime, incur higher repair costs and loss of revenue. These failures can belong to electrical, thermal or mechanical fault categories. Early detection of all these faults is very much critical for avoiding complete failure of these machines and requires continuous 24/7 online monitoring using sensors or intelligent electronic devices. However, an affordable low cost and efficient online monitoring is a desire to practice specifically for Medium Voltage (MV) machines which has a larger install base. Electrical Signature Analysis (ESA) technology offers such flexibility and requires measuring just current and/or voltage at a motor control panel for machine health diagnosis, unlike Vibration Analysis (VA) requiring installation of sensors and its wiring on the machine. Further, an intelligent, self-reliable and autonomous ESA procedure is required for monitoring the machines due to the lack of ESA standards in practice unlike for VA. This paper proposes a new Autonomous Electrical Signature Analysis (AESA) based measurement technique implemented in Intelligent Electronic Device (IED) i.e. Protection Relay offering 24/7 online monitoring. The proposed technique implemented in Relay not only avoids dependency on standalone ESA hardware but also provides earlier detection of failures using peak and energy magnitudes computation approach at fault frequencies. To validate the proposed method, various tests were performed on the actual 1000 HP and 300 HP motors with/without mechanical faults in a machine repair shop and results are discussed. Performance of the proposed method is also compared with commercially available third-party ESA device results proving the efficacy.
Intelligent data-driven machinery fault diagnosis methods have been successfully and popularly developed in the past years. While promising diagnostic performance has been achieved, the existing methods generally require large amounts of high-quality supervised data for training, which are mostly difficult and expensive to collect in real industries. Therefore, it is motivated that the distributed data of multiple clients can be integrated and exploited to build a powerful data-driven model. However, that basically requires data sharing among different users, and is not preferred in most industrial cases due to potential conflict of interests. In order to address the data island problem, a federated learning method for machinery fault diagnosis is proposed in this paper. Model training is locally implemented within each participated client, and a self-supervised learning scheme is proposed to enhance the learning performance. The server aggregates the locally updated models in each training round under the dynamic validation scheme, and a global fault diagnosis model can be established. Only the models are mutually communicated rather than the data, which ensures data privacy among different clients. The experiments on two datasets suggest the proposed method offers a promising approach on confidential decentralized learning.
Inter-turn fault (ITF), a common fault type of induction motors, can impact the safe and stable operation of the motor. Variations of magnetic field distribution and short circuit (SC) current were analyzed, and a diagnosis method of ITF was proposed based on the stator current complex component. Firstly, a 1.1 kW squirrel cage induction motor whose stator winding coils were rewound was taken as a prototype. The distributions of magnetic fields before and after the fault were obtained by calculating the finite element model. Based on the simulation and experimental results under no-load, light load, and rated load conditions, the phase current and SC loop current variations were obtained under different fault degrees. Then, the variations of current amplitudes in the faulty phase, non-faulty phases, and SC loop were analyzed to determine the impact on the operation of the motor caused by the fault. Finally, a diagnosis method of ITF, which uses the current complex component to represent the current variation, was proposed based on the calculation and analysis of faulty-motor current. The results show that this method can diagnose and locate faults effectively, verify the robustness of load variations, and provide a reference for the research on ITF of induction motors.
To combine the advantages of both model-driven and data-driven methods, this paper proposes a model-data-hybrid-driven (MDHD) method to diagnose open-switch faults in power converters. This idea is based on the explicit analytical model of converters and the learning capability of artificial neural network (ANN). The process of the method is divided into two parts: offline model analysis and learning, and online fault diagnosis. For both parts, model-driven and data-driven are combined. With the model information and data-based learning capability, a fast diagnosis for various operating conditions can be achieved without high computation burden, tricky threshold selection and complex rulemaking. This can greatly contribute to the practical application. The open-switch fault diagnosis in a two-level three-phase converter is studied for method validation. For this converter, an ANN is trained with two input elements, seven output elements, and two neurons in the hidden layer. Experimental results are given to demonstrate good performance.
Automatic and reliable fault diagnosis of rotating machinery cross working conditions is of practical importance. For this purpose, ensemble transfer convolutional neural networks (CNNs) driven by multi-channel signals are proposed in this paper. Firstly, a series of source CNNs modified with stochastic pooling and Leaky rectified linear unit (LReLU) are pre-trained using multi-channel signals. Secondly, the learned parameter knowledge of each individual source CNN is transferred to initialize the corresponding target CNN which is then fine-tuned by a few target training samples. Finally, a new decision fusion strategy is designed to flexibly fuse each individual target CNN to obtain the comprehensive result. The proposed method is used to analyze multi-channel signals measured from rotating machinery. The comparison result shows the superiorities of the proposed method over the existing deep transfer learning methods.