Neural networks application for induction motor faults diagnosis

Kowalski, C. T.; Orłowska-Kowalska, Teresa

doi:10.1016/s0378-4754(03)00087-9

Cited by 138 publications

(59 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Time-domain [4], frequency-domain [5][6][7][8], enhanced frequency [9][10][11][12], and time-scale analysis [13][14][15][16] are four main areas where signal processing techniques are used in the feature extraction [17]. The extracted features are used to both train and operate Artificial Neural Networks (ANNs) or other decision support systems (DSS) [18][19][20][21][22][23][24][25][26][27][28][29][30][31]. Extracting fixed features each time data is analyzed by DSS may require significant amount of computational effort.…”

Section: Introductionmentioning

confidence: 99%

Bearing Fault Detection by One‐Dimensional Convolutional Neural Networks

Eren

2017

Mathematical Problems in Engineering

186

106

View full text Add to dashboard Cite

Bearing faults are the biggest single source of motor failures. Artificial Neural Networks (ANNs) and other decision support systems are widely used for early detection of bearing faults. The typical decision support systems require feature extraction and classification as two distinct phases. Extracting fixed features each time may require a significant computational cost preventing their use in realtime applications. Furthermore, the selected features for the classification phase may not represent the most optimal choice. In this paper, the use of 1D Convolutional Neural Networks (CNNs) is proposed for a fast and accurate bearing fault detection system. The feature extraction and classification phases of the bearing fault detection are combined into a single learning body with the implementation of 1D CNN. The raw vibration data (signal) is fed into the proposed system as input eliminating the need for running a separate feature extraction algorithm each time vibration data is analyzed for classification. Implementation of 1D CNNs results in more efficient systems in terms of computational complexity. The classification performance of the proposed system with real bearing data demonstrates that the reduced computational complexity is achieved without a compromise in fault detection accuracy.

show abstract

Section: Introductionmentioning

confidence: 99%

Bearing Fault Detection by One‐Dimensional Convolutional Neural Networks

Eren

2017

Mathematical Problems in Engineering

186

106

View full text Add to dashboard Cite

show abstract

“…Even if experts are available, the technical information needed by the engineers is not always to hand, or received in the first instance. With the development of artificial intelligence techniques, many intelligent systems have been employed to assist the condition monitoring task to correctly interpret the fault data, such as expert systems, artificial neural networks (ANNs), support vector machines and fuzzy logic systems, and the results are promising (8) - (10) . Most efforts in the condition monitoring area have largely ignored the feature selection problem and have focused mainly on developing effective feature extraction methods and employing powerful clas-sifiers.…”

Section: Introductionmentioning

confidence: 99%

Fault Diagnosis System of Induction Motors Using Feature Extraction, Feature Selection and Classification Algorithm

Yang

Han

Yin

2006

JSME Int. J., Ser. C

View full text Add to dashboard Cite

This paper proposes a fault diagnosis system for induction motor which integrates principal component analysis (PCA), genetic algorithm (GA) and artificial neural network (ANN). Vibration signals and stator current signals are measured as the fault diagnosis media. Many sensors result in many features to ANN. In order to avoid the curse of dimensionality phenomenon and improve the classification rate, PCA and GA are employed to reduce the feature dimensionality of the measured data. PCA removes the relative features. Then the irrelative features after PCA are selected by GA to find better feature subset as inputs to the network under a few population and generations. GA is also used to optimize the ANN structure in that the selected PCs feature subset is evaluated by it. The efficiency of the proposed system is validated by comparison of other three systems: ANN only, ANN with PCA and ANN with GA. The classification success rate for the ANN with PCA and GA was 100% for validation, while the rates of ANN only, ANN with PCA and ANN with GA were 83.33%, 86.67% and 98.89%, respectively.

show abstract

“…Long time disturbances in technological processes cause big economic loses. The importance of incipient condition monitoring is a method of cost saving which is realized by detecting potential motor failures before they occur [1]. For this reason, the problem of fast fault detection and location as well as the problem of technical state evaluation are very significant in the industrial practice [2,3].…”

Section: Introductionmentioning

confidence: 99%

A method for condition monitoring and fault diagnosis in electromechanical system

Guo

Huang

et al. 2007

Neural Comput & Applic

View full text Add to dashboard Cite

Condition monitoring of electrical machines has received considerable attention in recent years. Many monitoring techniques have been proposed for electrical machine fault detection and localization. In this paper, the feasibility of using a nonlinear feature extraction method noted as Kernel independent component analysis (KICA) is studied and it is applied in self-organizing map to classify the faults of induction motor. In nonlinear feature extraction, we employed independent component analysis (ICA) procedure and adopted the kernel trick to nonlinearly map the Gaussian chirplet distributions into a feature space. First, the adaptive Gaussian chirplet distributions are mapped into an implicit feature space by the kernel trick, and then ICA is performed to extract nonlinear independent components of the Gaussian chirplet distributions. A thorough laboratory study shows that the diagnostic methods provide accurate diagnosis, high sensitivity with respect to faults, and good diagnostic resolution.

show abstract

Neural networks application for induction motor faults diagnosis

Cited by 138 publications

References 7 publications

Bearing Fault Detection by One‐Dimensional Convolutional Neural Networks

Bearing Fault Detection by One‐Dimensional Convolutional Neural Networks

Fault Diagnosis System of Induction Motors Using Feature Extraction, Feature Selection and Classification Algorithm

A method for condition monitoring and fault diagnosis in electromechanical system

Contact Info

Product

Resources

About