30 January 2018 in IEEE Transactions on Power Delivery
HNHebbale Narasimhaiah Nagamani
Hebbale Narasimhaiah Nagamani
The behavior of partial discharge (PD) acoustic emission (AE) signals in the transformer oil for a temperature range of 30°C to 75°C is studied via laboratory experiments. The AE signals are analyzed using the Fast Fourier Transform (FFT) to obtain the AE amplitude and the peak frequency. A peculiar behavior of reduction in the AE amplitude and the peak frequency observed above 65°C is reported. An attempt is made to explain the reduction in the AE amplitude after a certain temperature by correlating it with the properties of the transformer oil such as the breakdown voltage (BDV) and the viscosity. This AE signal analysis is perhaps an important advancement in the PD diagnosis and the PD spectrum analysis at different temperatures in transformer oil.
13 January 2025 in IEEE Transactions on Instrumentation and Measurement
Acoustic emission detection is a non-intrusive technique for monitoring transformer conditions by capturing emissions from partial discharges (PD), hotspots, and noise. Machine learning (ML) has been widely used in PD diagnostics in power transformers. While effective in managing complex data characteristics, traditional ML algorithms cannot quantify uncertainty in classification. This study uses Bayesian networks integrated with a classification algorithm to quantify uncertainty in acoustic signal classification. Bayesian deep learning (BDL) and ensemble models are the two commonly used techniques for uncertainty quantification. BDL models with multiple Bayesian layers are more prone to convergence issues with difficulty interpreting the sources of uncertainty. The performance of the ensemble method is based on model diversity with a higher risk of overfitting and does not offer insight into aleatoric uncertainty. The present work proposes architectures with a single Bayesian layer to quantify both aleatoric and epistemic uncertainties in acoustic signal classification. A Bayesian CNN layer combined with an ensemble architecture demonstrates comparatively higher performance among the considered Bayesian architectures. Raw acoustic signals are transformed into spectrogram images to enhance feature representation, capturing time-frequency characteristics. The proposed method is assessed using laboratory and field-measured data, demonstrating significant improvements in estimating uncertainty in classification.
6 March 2025 in IEEE Transactions on Plasma Science
Condition monitoring of the power transformer is important in extending the lifetime of the transformer through timely maintenance. Acoustic-emission (AE)-based partial discharge (PD) monitoring in a transformer is a commonly used technique for assessing the PD levels inside the transformer. The classification of the type of PD or localization of PD generally involves using multiple acoustic sensors, and recently, deep learning techniques have been widely used with the acquired PD data for classification. In the present work, a single-channel synchrosqueezing transform (SST)-based FastIVA blind-source-separation (BSS) technique in the time-frequency (TF) domain is proposed for the separation of multiple PD signals. SST, along with singular value decomposition (SVD), is used to identify the number of sources. The SST-FastIVA-based method is used for PD signal separation for the first time and it performs better than other separation techniques in noisy conditions. An SVM-based machine learning model is developed with original unmixed source signals for a classification task. The separated signals are tested using the developed model, and it attains an F1 score of 78.70% on the separated signals. Furthermore, the proposed methodology is used with the experimental data, and the F1 score obtained with the separated signal is 84.44%. The proposed single-channel (sensor) technique with BSS is a cost-effective method that can be used for the classification of multiple PD sources.
1 January 2016 in Serbian Journal of Electrical Engineering
HNHebbale Nagamani Narasimhaiah
Hebbale Nagamani Narasimhaiah
In power transformers, locating the partial discharge (PD) source is as important as identifying it. Acoustic Emission (AE) sensing offers a good solution for both PD detection and PD source location identification. In this paper the principle of the AE technique, along with in-situ findings of the online acoustic emission signals captured from partial discharges on a number of Generator Transformers (GT), is discussed. Of the two cases discussed, the first deals with Acoustic Emission Partial Discharge (AEPD) tests on two identical transformers, and the second deals with the AEPD measurement of a transformer carried out on different occasions (years). These transformers are from a hydropower station and a thermal power station in India. Tests conducted in identical transformers give the provision for comparing AE signal amplitudes from the two transformers. These case studies also help in comprehending the efficacy of integrating Dissolved Gas is (DGA) data with AEPD test results in detecting and locating the PD source.
1 January 2024 in Serbian Journal of Electrical Engineering
Signal processing helps monitor the condition of power equipment. Partial discharge (PD) signals used in condition-based maintenance give crucial information in the diagnosis of degradation of insulation. The acoustic emission technique (AET) is one of the most widely used techniques in PD signal analysis due to its inherent advantages. Analyzing acoustic emission partial discharge (AEPD) signals in the wavelet-domain provides critical insights into the location and type of the sources of PD. Selection of the most suitable mother wavelet in applying discrete wavelet transform (DWT) on AEPD signals is important as it will directly impact the outcome. For this selection, 36 wavelets belonging to the Daubechies, Symlets, Coiflets, and Bi-orthogonal families are investigated. For this purpose, five experimentally collected AEPD test signals are used. The selection is based on the ?accuracy of wavelet decomposition results? in this work, probably for the first time. One mother wavelet from each family is individually shortlisted for all three performances, namely (a) reconstruction, (b) denoising, and (c) compression, by computing and comparing their commonly used metrics. Further, based on percentage energy criteria, the most suitable mother wavelets are identified as coif3, coif4, and coif5, respectively, for the three performances.
1 October 2023 in IEEE Transactions on Dielectrics and Electrical Insulation
The study describes the use of nano aluminum nitride in developing thermally conductive and electrically insulating materials. An experimental approach is followed, starting with the hydrophobic surface treatment of nanofillers using HMDS. TEM and XRD analysis are used to confirm the nanofiller presence and gain insight regarding the filler shape, dimensions, and phase. EDX, FTIR and XRD analysis revealed the co-existence of aluminum oxide in traces along with aluminum nitride. XRD patterns revealed no change after surface modification and confirmed the presence of an amorphous functionalized layer. The protocol for composite compounding is described in detail. SEM analysis of composites revealed a moderate degree of dispersion with localized pockets of agglomeration. Further, the TEM analysis performed gave insight into the dispersion of nanofillers. Finally, DSC analysis is performed to understand the effects of nano aluminum nitride on glass transition, melting point, and crystallization temperatures. An interesting pattern of increase in glass transition temperatures with nanofiller loading levels is discussed, drawing attention to the aggregated filler effect and the disentangling of silicone chains with increasing interfacial interaction. TGA analysis revealed an early onset of degradation with increase in nano aluminum nitride loading levels. Thermal conductivity increased with nanofiller loading levels, peaking at 3 phr by weight. Dielectric studies of breakdown strength and frequency domain spectroscopy from 1 mHz to 1 KHz is studied under different thermal points. The results are presented and discussed here.