泛研网

公众版 查看说明
全球科研项目数据库2026版

Development of methods for fault diagnosing of an induction motor using current signals under conditions of a limited data set

项目来源

俄罗斯科学基金(RSF)

项目主持人

Ibryaeva Olga

项目受资助机构

Federal State Autonomous Educational Institution of Higher Education"South Ural State University"

立项年度

2025

立项时间

未公开

项目编号

25-29-00633

项目级别

国家级

研究期限

未知 / 未知

受资助金额

未知

学科

ENGINEERING SCIENCES-Integration and information processing in technical systems

学科代码

09-09-605

基金类别

未公开

мониторинг состояния ; диагностика неисправностей ; классификация дефектов ; искусственный интеллект ; машинное обучение ; нейронные сети ; смешанные входные данные ; выделение признаков ; Преобразование Гильберта-Хуанга ; вейвлет-преобразование ; GAF ; MTF ; RP ; аугментация данных ; condition monitoring ; fault diagnosis ; defect classification ; artificial intelligence ; machine learning ; neural networks ; mixed input data ; feature extraction ; Hilbert-Huang transform ; wavelet transform ; data augmentation

参与者

未公开

参与机构

未公开

项目标书摘要：nnotation:Minimizing unscheduled failures and equipment downtime is a key point in increasing the economic efficiency of industrial production.The induction motor ranks first among the types of electric machines used;most variable-frequency electric drives are built on the basis of induction motors.According to various estimates,10-15%of the entire motor fleet fails per year.The main causes of motor failures in electric drives are associated with increased loads and untimely maintenance.Fault detection methods developed to date show high efficiency in laboratory installations under stationary operating conditions.However,in industrial environments,motors often change speed and load,which leads to a significant deterioration in the accuracy of diagnostic models that are not adapted to unsteady operating conditions.Data corresponding to the operation of a defective motor is often unavailable or available in limited quantities.In addition,the installation of additional sensors and any interference in the technological process are,as a rule,unacceptable.Most suitable for industrial environments,diagnostics are based on current signals that can be easily and non-invasively collected from the motor.In this regard,this study is aimed at solving the problem of diagnosing the state of an induction motor using current signals under conditions of a limited set of data and non-stationary operating modes.It is necessary to develop methods capable of determining the condition of the motor when its operating conditions change,noise and the“masking”effect of the technological process,control using a frequency converter,as well as partial or complete absence of data on the defective state.The research will develop a new method for augmenting current signal data from the motor,taken at certain values of load and supply frequency,to other operating modes.The method will take into account the features of the spectra of current signals under different motor operating modes and use their artificial distortion.This expansion of the training set with synthetic data will further allow us to create a neural network diagnostic model capable of operating in a non-stationary mode.The model will be trained using images obtained from current signals,which is also an innovative approach.To develop a hybrid motor diagnostic model,a mixed input from a vector of numerical features and images will be used for the first time.Such simultaneous recording of different types of data by the network will ensure its greater accuracy.The project team has a scientific background in methods of obtaining and processing information(the team’s experience in the field of signal processing is more than 10 years),as well as successful projects in the field of technical diagnostics and measuring equipment(RFBR,UMNIK program,state task of the Ministry of Education,Priority-2030 program,in within the framework of the strategic project“Intelligent Manufacturing”),successful R&D under the agreement with PJSC MMK.Expected results:During the work on the project the following results will be obtained:1)Features of the spectra of current signals corresponding to different states and operating modes of the induction motor have been identified.2)A new method has been developed for augmenting current signals from the motor,obtained at certain values of load and supply frequency,to other operating modes using spectrum distortion.3)Methods for converting synthetic and real current signals into GAF,MTF,RP,HHT,WT images have been studied.4)An intelligent method for diagnosing faults in an induction motor has been developed based on a convolutional neural network and a limited set of training data corresponding to the case of operation of a defective motor,including in the complete absence of such data.5)Informative numerical statistical features from current signals are identified.6)A hybrid motor diagnostic model has been developed with mixed input from different types of data:images and numerical features extracted from signals from the motor.The scientific result of the project will be the development of methods for obtaining and processing data,methods for identifying informative features,artificial intelligence methods for solving the problem of detecting a defect and its type.

Application Abstract: Annotation:Minimizing unscheduled failures and equipment downtime is a key point in increasing the economic efficiency of industrial production.The induction motor ranks first among the types of electric machines used;most variable-frequency electric drives are built on the basis of induction motors.According to various estimates,10-15%of the entire motor fleet fails per year.The main causes of motor failures in electric drives are associated with increased loads and untimely maintenance.Fault detection methods developed to date show high efficiency in laboratory installations under stationary operating conditions.However,in industrial environments,motors often change speed and load,which leads to a significant deterioration in the accuracy of diagnostic models that are not adapted to unsteady operating conditions.Data corresponding to the operation of a defective motor is often unavailable or available in limited quantities.In addition,the installation of additional sensors and any interference in the technological process are,as a rule,unacceptable.Most suitable for industrial environments,diagnostics are based on current signals that can be easily and non-invasively collected from the motor.In this regard,this study is aimed at solving the problem of diagnosing the state of an induction motor using current signals under conditions of a limited set of data and non-stationary operating modes.It is necessary to develop methods capable of determining the condition of the motor when its operating conditions change,noise and the“masking”effect of the technological process,control using a frequency converter,as well as partial or complete absence of data on the defective state.The research will develop a new method for augmenting current signal data from the motor,taken at certain values of load and supply frequency,to other operating modes.The method will take into account the features of the spectra of current signals under different motor operating modes and use their artificial distortion.This expansion of the training set with synthetic data will further allow us to create a neural network diagnostic model capable of operating in a non-stationary mode.The model will be trained using images obtained from current signals,which is also an innovative approach.To develop a hybrid motor diagnostic model,a mixed input from a vector of numerical features and images will be used for the first time.Such simultaneous recording of different types of data by the network will ensure its greater accuracy.The project team has a scientific background in methods of obtaining and processing information(the team’s experience in the field of signal processing is more than 10 years),as well as successful projects in the field of technical diagnostics and measuring equipment(RFBR,UMNIK program,state task of the Ministry of Education,Priority-2030 program,in within the framework of the strategic project“Intelligent Manufacturing”),successful R&D under the agreement with PJSC MMK.Expected results:During the work on the project the following results will be obtained:1)Features of the spectra of current signals corresponding to different states and operating modes of the induction motor have been identified.2)A new method has been developed for augmenting current signals from the motor,obtained at certain values of load and supply frequency,to other operating modes using spectrum distortion.3)Methods for converting synthetic and real current signals into GAF,MTF,RP,HHT,WT images have been studied.4)An intelligent method for diagnosing faults in an induction motor has been developed based on a convolutional neural network and a limited set of training data corresponding to the case of operation of a defective motor,including in the complete absence of such data.5)Informative numerical statistical features from current signals are identified.6)A hybrid motor diagnostic model has been developed with mixed input from different types of data:images and numerical features extracted from signals from the motor.The scientific result of the project will be the development of methods for obtaining and processing data,methods for identifying informative features,artificial intelligence methods for solving the problem of detecting a defect and its type.

  • 排序方式：
  • 1
  • /

  • 1.Hybrid CNN–MLP for Robust Fault Diagnosis in Induction Motors Using Physics-Guided Spectral Augmentation

    • 关键词：
    • Convolution;Convolutional neural networks;Electric fault currents;Failure analysis;Fault detection;Multilayer neural networks;Multilayers;Broken rotor bar;Convolutional neural network;Data augmentation;Diagnostic model;Faults diagnosis;Inductions motors;Multilayers perceptrons;Operating regimes;Spectral warping;Warpings
    • Shestakov, Alexander;Galyshev, Dmitry;Ibryaeva, Olga;Eremeeva, Victoria
    • 《Algorithms》
    • 2025年
    • 18卷
    • 11期
    • 期刊

    The diagnosis of faults in induction motors, such as broken rotor bars, is critical for preventing costly emergency shutdowns and production losses. The complexity of this task lies in the diversity of induction motor operating regimes. Specifically, a change in load alters the signal’s frequency composition and, consequently, the values of fault diagnostic features. Developing a reliable diagnostic model requires data covering the entire range of motor loads, but the volume of available experimental data is often limited. This work investigates a data augmentation method based on the physical relationship between the frequency content of diagnostic signals and the motor’s operating regime. The method enables stretching and compression of the signal in the spectral domain while preserving Fourier transform symmetry and energy consistency, facilitating the generation of synthetic data for various load regimes. We evaluated the method on experimental data from a 0.37 kW induction motor with broken rotor bars. The synthetic data were used to train three diagnostic models: a Multilayer Perceptron (MLP), a Convolutional Neural Network (CNN), and a hybrid CNN-MLP model. Results indicate that the proposed augmentation method enhances classification quality across different load levels. The hybrid CNN-MLP model achieved the best performance, with an F1-score of 0.98 when augmentation was employed. These findings demonstrate the practical efficacy of physics-guided spectral augmentation for induction motor fault diagnosis. © 2025 by the authors.

    ...

  • 2.Broken Rotor Bar Fault Diagnosis Using Spectrum Distortion for Data Augmentation

    • 关键词：
    • Convolution;Discrete wavelet transforms;Image analysis;Image coding;Image compression;Nonlinear distortion;Broken rotor bar;Convolutional neural network;Current signal;Data augmentation;Hilbert transform;Inductions motors;Operating modes;Scalogram;Spectrum distortion;Wavelets transform
    • Shestakov, Alexander L.;Galyshev, Dmitrii V.;Eremeeva, Victoria A.;Ibryaeva, Olga L.
    • 《27th International Conference on Digital Signal Processing and its Applications, DSPA 2025》
    • 2025年
    • March 26, 2025 - March 28, 2025
    • Moscow, Russia
    • 会议

    The problem of diagnostics of broken rotor bars in an induction (IM) motor based on its current signals is considered. The proposed fault detection technique is based on filtering the current signal in the region around the frequency 7f, where f is the power supply frequency, and further obtaining its envelope using the Hilbert Transform. Scalograms are constructed from the envelopes using a Continuous Wavelet Transform and form a training dataset for the Convolutional Neural Network (CNN) classifier. In real industrial conditions, we have a lack of defective data which is simulated in this paper. It is shown that a CNN model trained on normal data obtained under different motor operating conditions and defective data obtained under one operating mode will poorly detect a defect when changing the operating mode. The reason for this lies in the characteristic defect frequencies change with the change in the operating mode. A data augmentation method based on spectrum distortion is proposed to solve this problem. We distort the spectrum of a real defective signal taking into account its structure and create synthetic data that simulate missing defective signals with different operating conditions. The paper presents the experimental rig and dataset, and shows that the spectral distortion-based augmentation technique significantly improves the performance of the CNN model under conditions of insufficient defective training data. © 2025 IEEE.

    ...
  • 排序方式：
  • 1
  • /