A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Classification and Authentication of Induction Motor Faults using Time and Frequency Feature Dependent Probabilistic Neural Network Model
A probabilistic neural network (PNN)-based robust classifier has been developed in this work for classification and authentication of induction motor faults using the features of three domains. The decomposition level of wavelet has also been selected to get highest classification accuracy, and the optimal value of PNN spread parameter has also been estimated in this work. The three-phase line currents are collected from induction motors for six classes of faults and one healthy condition under three different loadings. Time domain, frequency domain and time–frequency domain features are analyzed here using direct fault current signals, their FFT spectrums and wavelet transform of the same, respectively, followed by principal component analysis (PCA). PNN is used to develop the final classifier using PCA features. Inclusion of variability in loading ensures robustness. The PNN model is further tuned with varying spread parameter to obtain the optimum level of accuracy, and the results are also recorded. The proposed model is found to detect faults with highest mean accuracy exceeding 99%. The classification accuracies, obtained using different schemes, are analyzed and compared. Besides, the proposed method incorporates low computational PNN architecture. High accuracy of fault classification combined with simplicity of analysis indicates its effectiveness for diagnosis of various induction motor faults as well as its ease of implementation in developing real-time condition monitoring and fault diagnosis schemes.
Classification and Authentication of Induction Motor Faults using Time and Frequency Feature Dependent Probabilistic Neural Network Model
A probabilistic neural network (PNN)-based robust classifier has been developed in this work for classification and authentication of induction motor faults using the features of three domains. The decomposition level of wavelet has also been selected to get highest classification accuracy, and the optimal value of PNN spread parameter has also been estimated in this work. The three-phase line currents are collected from induction motors for six classes of faults and one healthy condition under three different loadings. Time domain, frequency domain and time–frequency domain features are analyzed here using direct fault current signals, their FFT spectrums and wavelet transform of the same, respectively, followed by principal component analysis (PCA). PNN is used to develop the final classifier using PCA features. Inclusion of variability in loading ensures robustness. The PNN model is further tuned with varying spread parameter to obtain the optimum level of accuracy, and the results are also recorded. The proposed model is found to detect faults with highest mean accuracy exceeding 99%. The classification accuracies, obtained using different schemes, are analyzed and compared. Besides, the proposed method incorporates low computational PNN architecture. High accuracy of fault classification combined with simplicity of analysis indicates its effectiveness for diagnosis of various induction motor faults as well as its ease of implementation in developing real-time condition monitoring and fault diagnosis schemes.
Classification and Authentication of Induction Motor Faults using Time and Frequency Feature Dependent Probabilistic Neural Network Model
J. Inst. Eng. India Ser. B
Thakur, Arunava Kabiraj (author) / Mukherjee, Alok (author) / Kundu, Palash Kumar (author) / Das, Arabinda (author)
Journal of The Institution of Engineers (India): Series B ; 104 ; 623-640
2023-06-01
18 pages
Article (Journal)
Electronic Resource
English
Neural Network Based Expert System for Induction Motor Faults Detection
| British Library Online Contents | 2006
When time and faults matter: towards a time-dependent probabilistic SHA in Calabria, Italy
| British Library Online Contents | 2017