Detection of inter-turn short circuits in induction motors using orthogonal matching pursuit and dictionary learning

Three-phase induction motors are widely adopted in industrial systems due to their robustness, ease of maintenance, and simple operation. However, they are prone to various types of faults, notably stator winding faults. Previous research indicates that 20–40% of three-phase induction motor failures are stator-related, with inter-turn short circuits as a leading cause. These faults can pose significant risks to both the motor and connected equipment. Therefore, the early detection of inter-turn short circuit (ITSC) faults is essential to prevent system breakdowns and improve the safety and reliability of industrial operations. This paper presents a comparative investigation of two distinct diagnostic methodologies for the detection of ITSC faults in induction motors. The first methodology is based on a Motor Current Signature Analysis (MCSA) utilizing the short-time Fourier transform (STFT) for the real-time monitoring of fault-related harmonics. The second methodology is centered around the monitoring of the zero-sequence voltage (ZSV). The findings from several experimental tests performed on a 1.1 kW three-phase induction motor across a range of operating conditions highlight the superior performance of the ZSV method with respect to the MCSA-based STFT method in terms of reliability, rapidity, and precision for the diagnosis of ITSC faults.

Inter-turn fault detection of induction motors using a method based on spectrogram of motor currents

In this paper, an experimental validation of an efficient approach to the Fault Detection and Isolation (FDI) of Induction Motor (IM) is proposed. The problem of Inter-turn short circuits (ITSC) in the stator windings is addressed. By introducing fault factors in the IM model an observer-based residual generator is designed, allowing the detection of ITSC in stator windings. The residual generator is built around an extended Kalman Filter (EKF) in order to estimate state variables and fault factors, which permits the evaluation of the severity of the fault. To overcome the problem of tuning the EKF a PSO algorithm is developed. It carries out a heuristic search of the noise matrices by optimizing a cost function. The proposed solution is validated by computer simulations and by real-time implementation on dSPACE 1104 Digital Signal Processor (DSP) test-bench under the healthy and the faulty conditions of IM. To perform tests under faulty conditions, an IM with customized design is built and the stator is rewound permitting to create ITSC. The results reveal the quick detection of the faults, the quantification of its severity and confirm the efficacy of this observer-based FDI algorithm.

In this paper, an experimental validation of an efficient approach to the Fault Detection and Isolation (FDI) of Induction Motor (IM) is proposed. The problem of Inter-turn short circuits (ITSC) in the stator windings is addressed. By introducing fault factors in the IM model an observer-based residual generator is designed, allowing the detection of ITSC in stator windings. The residual generator is built around an extended Kalman Filter (EKF) in order to estimate state variables and fault factors, which permits the evaluation of the severity of the fault. To overcome the problem of tuning the EKF a Partcile swarm Optimisation (PSO) algorithm is developed. It carries out a heuristic search of the noise matrices by optimizing a cost function. The proposed solution is validated by computer simulations and by real-time implementation on dSPACE 1104 Digital Signal Processor (DSP) test-bench under the healthy and the faulty conditions of IM. To perform tests under faulty conditions, an IM with customized design is built and the stator is rewound permitting to create ITSC. The results reveal the quick detection of the faults, the quantification of its severity and confirm the efficacy of this observer-based FDI algorithm.

This work presents a lightweight and practical methodology for detecting inter-turn short-circuit faults in squirrel-cage induction motors under different mechanical load conditions. The proposed approach utilizes a one-dimensional convolutional neural network (1D-CNN) enhanced with residual blocks and trained on differentiated stator current signals obtained under different load mechanical conditions. This preprocessing step enhances fault-related features, enabling improved learning while maintaining the simplicity of a lightweight CNN. The model achieved classification accuracies above 99.16% across all folds in five-fold cross-validation and demonstrated the ability to detect faults involving as few as three short-circuited turns. Comparative experiments with the Multi-Scale 1D-ResNet demonstrate that the proposed method achieves similar or superior performance while significantly reducing training time. These results highlight the model’s suitability for real-time fault detection in embedded and resource-constrained industrial environments.

This work proposes an alternative strategy to the use of a speed sensor in the implementation of active and reactive power based model reference adaptive system (PQ-MRAS) estimator in order to calculate the rotor and stator resistances of an induction motor (IM) and the use of these parameters for the detection of inter-turn short circuits (ITSC) faults in the stator of this motor. The rotor and stator resistance estimation part of the IM is performed by the PQ-MRAS method in which the rotor angular velocity is reconstructed from the interconnected high gain observer (IHGO). The ITSC fault detection part is done by the derivation of stator resistance estimated by the PQ-MRAS estimator. In addition to the speed sensorless detection of ITSC faults of the IM, an approach to determine the number of shorted turns based on the difference between the phase current of the healthy and faulty machine is proposed. Simulation results obtained from the MATLAB/Simulink platform have shown that the PQ-MRAS estimator using an interconnected high-gain observer gives very similar results to those using the speed sensor. The estimation errors in the cases of speed variation and load torque are almost identical. Variations in stator and rotor resistances influence the performance of the observer and lead to poor estimation of the rotor resistance. The results of ITSC fault detection using IHGO are very similar to the results in the literature using the same diagnostic approach with a speed sensor.

Fault detection in induction motors operating in non-stationary regimes has become a need in today’s industry. Most of the works published deal with line-fed motors. Nevertheless, the number of inverter-fed induction motors has significantly increased in recent years. Therefore, several fault detection techniques have been proposed lately for this type of motors, based mainly on an adequate input signal processing to obtain fault signatures in the time-frequency domain. In this paper, a comparison of time-frequency techniques applied to fault detection in inverter-fed induction motors in a transient state is presented. For that purpose, the techniques are applied to two current signals acquired from two induction motors with two types of faults: bar breakage and mixed eccentricity. The paper shows the particularities and special difficulties of diagnosing under this type of feeding, reviewing the works related to each technique. The strengths and weaknesses of these techniques are discussed with the goal of providing a criterion for its application in an industrial environment and guidance for future developments in this field.

Induction motor failures are mainly due to stator and rotor faults. In this paper, a novel method based on Park's vector approach for fault detection of induction motor is presented. Using suggested method it is possible to detect the various types of faults in squirrel induction motors. Further, paper focuses on the detection and discrimination of the simultaneous stator inter turn and broken rotor bar faults in induction motor. In order to evaluate the ability of the proposed method several experimental results are presented. The result demonstrates that the proposed method is effective and accurate and can be widely used in the induction motor fault detection.

This paper examines the detection of mechanical faults in induction motors by stator current analysis. Mechanical faults lead generally to periodic load torque oscillations. The influence of the torque oscillations on the induction motor stator current is studied. The mechanical fault results in a sinusoidal phase modulations of the stator current. Based on these assumptions, several signal processing methods suitable for stator current signature analysis are discussed: classical spectral analysis, instantaneous frequency estimation and time-frequency analysis using the Wigner distribution. Experimental and simulation results validate the theoretical approach

This work studies the acoustic imaging problem with compressed sensing (CS) and microphone arrays. The CS algorithm with Basis Pursuit (BP) algorithm has shown satisfying results in acoustic imaging, the maps of which are characterized by super-resolution. However, the performance of the CS algorithm with the BP algorithm is limited to Restricted Isometry Property (RIP), and the algorithm has a long CPU-time. We propose a new CS algorithm with Orthogonal Matching Pursuit (OMP) algorithm for acoustic imaging. The performance of the OMP algorithm with regard to RIP is examined through numerical simulation in this work. The simulation results and CPU-time for OMP algorithm are compared with those of the BP algorithm and the conventional beamformer (CBF). When the RIP does not hold, satisfying results can still be obtained by the OMP algorithm, and the CPU-time for OMP algorithm is far less than BP algorithm. In order to validate the feasibility of the OMP algorithm in acoustic imaging, an experiment is also conducted in a semi-anechoic room. Two mobile phones are served as sound sources. We investigate the mobile phones sources and compare the experimental results with those of BP algorithm and CBF method. The OMP algorithm can locate the main sources at low frequencies, while the CBF method can just give a rough indication and fails for low frequencies due to the width of its main lobe. Due to many reconstructed sources outside of the expected source positions existing on the map, the BP algorithm fails to locate the main sources at low frequencies.

Due to the importance of induction motors in a wide variety of industrial processes, it is crucial to properly identify abnormal conditions in order to avoid unexpected stops. The inter-turn short circuit (ITSC) is a very common failure produced with electrical stresses and affects induction motors (IMs), leading to catastrophic damage. Therefore, this work proposes the use of the empirical wavelet transform to characterize the time frequency behavior of the IM combined with a self-organizing map (SOM) structure to perform an automatic detection and classification of different severities of ITSC. Since the amount of information obtained from the empirical wavelet transform is big, a genetic algorithm is implemented to select the modes that allow a reduction in the quantization error in the SOM. The proposed methodology is applied to a real IM during the start-up transient considering four different fundamental frequencies. The results prove that this technique is able to detect and classify three different fault severities regardless of the operation frequency.

Electric motors play a fundamental role in various industries, and their relevance is strengthened in the context of the energy transition. Having efficient tools and techniques to detect and diagnose faults in electrical machines is crucial, as is providing early alerts to facilitate prompt decision-making. This study proposes indicators based on the magnitude of the space vector stator current for detecting and diagnosing incipient inter-turn short circuits (ITSCs) in induction motors (IMs). The effectiveness of these indicators was evaluated using four machine learning methods previously documented in the literature: random forests (RFs), support vector machines (SVMs), the k-nearest neighbor (kNN), and feedforward and recurrent neural networks (FNNs and RNNs). This assessment was conducted using experimental data. The results were compared with indicators based on discrete wavelet transform (DWT), demonstrating the viability of the proposed approach, which opens up a way of detecting incipient ITSCs in three-phase IMs. Furthermore, utilizing features derived from the magnitude of the spatial vector led to the successful identification of the phase affected by the fault.

This paper present the stator inter-turn short circuit fault detection in induction machine from stator current analysis using two techniques, such as the fast Fourier transform (FFT), which utilize the steady-state spectral components of the stator quantities. The accuracy of these techniques depends on the load and speed of the machine and the discrete wavelet transform (DWT) witch considered an ideal tool for this purpose, due to its suitability for the analysis of signal whose frequency spectrum is variable in time. So, it will study the problem of stator inter-turn short circuit. Simulation and experimentally results are presented in order to compare the performances of these techniques for inter-turn short circuit fault diagnosis.

Aiming at the problems of complex data processing and difficult identification caused by noise interference and other factors in the process of fault type identification of distribution network, this paper proposes a fault type identification method of distribution network based on sparse representation. The collection of fault data does not rely on the Shannon Theory. The collected time domain three-phase voltage(TPV) and zero-sequence voltage(ZSV) are divided into training sets and test sets. Learn the characteristic information of the fault voltage of the training set by the K-SVD dictionary learning algorithm, and construct an over-complete dictionary that accurately matches the essential characteristics of various types of faults. Based on learning dictionaries, the orthogonal matching pursuit (OMP) algorithm is used to perform adaptive dictionary sparse decomposition of the original time-domain fault data of the test set, and the fault signal is decomposed into the product of the overcomplete dictionary and the sparse vector. Combined with the classification method based on sparse Representation(SRC), the classification of the reconstructed fault signal is achieved. The proposed fault classification method does not need to construct fault signal features manually and avoids the tedious process of fault signal feature screening. It has strong adaptability and is not susceptible to transition resistance and noise. MATLAB/Simulink simulation verifies the effectiveness and accuracy of the proposed method.

The orthogonal matching pursuit (OMP) algorithm is widely used because of its simple algorithmic structure, the linear magnitude of its computational complexity and other advantages. Algorithms of improved and extended OMP types are continually emerging. To make the OMP algorithm more suitable for the new transmission requirements of the fifth-generation (5G) network, the orthogonal matching pursuit with multiple users (OMP_mu) algorithm is proposed for cell-free massive multi-input multi-output (MIMO) networks with multiuser spatial index modulation. The OMP_mu algorithm considers the constraint condition for the sparsity of each user on the basis of each iterative search in the OMP algorithm until all users meet the constraint. The simulation results show that compared with the original OMP algorithm, the OMP_mu algorithm has a greatly improved demodulation performance. Compared with the orthogonal matching pursuit step-by-step (OMP_step) algorithm, which takes the user sparsity constraint condition into account in two steps, the OMP_mu algorithm greatly reduces the operation time and number of iterations while achieving a slightly better performance. The OMP_mu algorithm provides an effective reference for the application of spatial index modulation in cell-free massive MIMO networks.