Tensor Based Algorithm for Automatic Partial Discharges Pattern Classification

The development of automated tools capable of monitoring electric motors is important for industrial applications. The measure of partial discharges is one of the most prominent methodologies for the evaluation of electric motors operating conditions. This work proposes to apply tensor-based classification methods to discriminate between different partial discharges sources. The major advantage of tensor-based approaches relies on the fact that they can inherently handle partial discharges pattern using all the geometrical information embedded in the patterns. Secondly, noise filtering can be embedded in classification mechanism, without manual tuning. As a consequence, better performances are obtained in presence of additive sources of noise like 4G, GSM connections and electromagnetic sources that could be detected by conducted and irradiated sensors especially during an online monitoring. To demonstrate the ability of the tensor-based methods to overcome the noise presence, tests involved defects with different levels of addictive noises. Experimental results proved the effectiveness of the proposed approach.

[1]  V. M. Catterson,et al.  Deep neural networks for understanding and diagnosing partial discharge data , 2014, 2014 International Conference on Advances in Communication and Computing Technologies (ICACACT 2014).

[2]  Paolo Gastaldo,et al.  A Digital Implementation of Extreme Learning Machines for Resource-Constrained Devices , 2018, IEEE Transactions on Circuits and Systems II: Express Briefs.

[3]  L. Hao,et al.  Partial discharge identification using a support vector machine , 2005, CEIDP '05. 2005 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, 2005..

[4]  Johan A. K. Suykens,et al.  Learning with tensors: a framework based on convex optimization and spectral regularization , 2014, Machine Learning.

[5]  Alex Smola,et al.  Kernel methods in machine learning , 2007, math/0701907.

[6]  A. Contin,et al.  Classification and separation of partial discharge signals by means of their auto-correlation function evaluation , 2009, IEEE Transactions on Dielectrics and Electrical Insulation.

[7]  Gene H. Golub,et al.  Singular value decomposition and least squares solutions , 1970, Milestones in Matrix Computation.

[8]  Venizelos Efthymiou,et al.  Feature extraction of partial discharge signals using the wavelet packet transform and classification with a probabilistic neural network , 2010 .

[9]  F. Guastavino,et al.  Hardware Friendly Neural Network for the PD Classification , 2018, 2018 IEEE Conference on Electrical Insulation and Dielectric Phenomena (CEIDP).

[10]  Gehao Sheng,et al.  GIS partial discharge pattern recognition via deep convolutional neural network under complex data source , 2018, IEEE Transactions on Dielectrics and Electrical Insulation.

[11]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[12]  C.S. Chang,et al.  Separation of corona using wavelet packet transform and neural network for detection of partial discharge in gas-insulated substations , 2005, IEEE Transactions on Power Delivery.

[13]  Maria-Florina Balcan,et al.  A theory of learning with similarity functions , 2008, Machine Learning.

[14]  L. Hao,et al.  Partial discharge source discrimination using a support vector machine , 2010, IEEE Transactions on Dielectrics and Electrical Insulation.

[15]  Paolo Gastaldo,et al.  Learning with Similarity Functions: a Tensor-Based Framework , 2018, Cogn. Comput..

[16]  Chengke Zhou,et al.  Application of K-Means method to pattern recognition in on-line cable partial discharge monitoring , 2013, IEEE Transactions on Dielectrics and Electrical Insulation.