TY - JOUR
T1 - Moisture estimation in power transformer oil using acoustic signals and spectral kurtosis
AU - Leite, Valéria C.M.N.
AU - Veloso, Giscard F.C.
AU - Borges da Silva, Luiz Eduardo
AU - Lambert-Torres, Germano
AU - Borges da Silva, Jonas G.
AU - Pinto, João Onofre Pereira
N1 - Publisher Copyright:
© 2016 IOP Publishing Ltd.
PY - 2016/1/14
Y1 - 2016/1/14
N2 - The aim of this paper is to present a new technique for estimating the contamination by moisture in power transformer insulating oil based on the spectral kurtosis analysis of the acoustic signals of partial discharges (PDs). Basically, in this approach, the spectral kurtosis of the PD acoustic signal is calculated and the correlation between its maximum value and the moisture percentage is explored to find a function that calculates the moisture percentage. The function can be easily implemented in DSP, FPGA, or any other type of embedded system for online moisture monitoring. To evaluate the proposed approach, an experiment is assembled with a piezoelectric sensor attached to a tank, which is filled with insulating oil samples contaminated by different levels of moisture. A device generating electrical discharges is submerged into the oil to simulate the occurrence of PDs. Detected acoustic signals are processed using fast kurtogram algorithm to extract spectral kurtosis values. The obtained data are used to find the fitting function that relates the water contamination to the maximum value of the spectral kurtosis. Experimental results show that the proposed method is suitable for online monitoring system of power transformers.
AB - The aim of this paper is to present a new technique for estimating the contamination by moisture in power transformer insulating oil based on the spectral kurtosis analysis of the acoustic signals of partial discharges (PDs). Basically, in this approach, the spectral kurtosis of the PD acoustic signal is calculated and the correlation between its maximum value and the moisture percentage is explored to find a function that calculates the moisture percentage. The function can be easily implemented in DSP, FPGA, or any other type of embedded system for online moisture monitoring. To evaluate the proposed approach, an experiment is assembled with a piezoelectric sensor attached to a tank, which is filled with insulating oil samples contaminated by different levels of moisture. A device generating electrical discharges is submerged into the oil to simulate the occurrence of PDs. Detected acoustic signals are processed using fast kurtogram algorithm to extract spectral kurtosis values. The obtained data are used to find the fitting function that relates the water contamination to the maximum value of the spectral kurtosis. Experimental results show that the proposed method is suitable for online monitoring system of power transformers.
KW - acoustic emission
KW - condition monitoring
KW - higher-order statistics
KW - moisture measurement
KW - oil insulation
KW - partial discharges
KW - power transformers
UR - http://www.scopus.com/inward/record.url?scp=84958225820&partnerID=8YFLogxK
U2 - 10.1088/0957-0233/27/3/035301
DO - 10.1088/0957-0233/27/3/035301
M3 - Article
AN - SCOPUS:84958225820
SN - 0957-0233
VL - 27
JO - Measurement Science and Technology
JF - Measurement Science and Technology
IS - 3
M1 - 035301
ER -
