M-robust algorithm for detecting power transformer inner faults accompanied by an electrical arc based on sound signal processing is presented. The electrical arc, which exists during the fault, generates specific sound waves. The efficiency of the presented algorithm for processing these types of sound signals is confirmed by performing experimental tests. It has been shown that even in the presence of different types of disturbances (including impulse disturbance), the M-robust methodology gives fast and reliable results which lead to the detection of transformer inner fault with electrical arc. The presented algorithm efficiently and quickly detects the unwanted processes in the transformer which do not result in a current increase (the initial discharge which precedes the electrical arc) and which could not be detected by the standard protection functions. The presented methodology gives an opportunity for further development and online implementation of a simple, low cost and broadly applicable system for the detection of internal faults in transformers, independently of the voltage and power level.
[1]
V. Miranda,et al.
Entropy and Correntropy Against Minimum Square Error in Offline and Online Three-Day Ahead Wind Power Forecasting
,
2009,
IEEE Transactions on Power Systems.
[2]
M. Boutayeb,et al.
Identification of the Induction Motor in Sinusoidal Mode
,
2010,
IEEE Transactions on Energy Conversion.
[3]
E. Kyriakides,et al.
PMU Measurement Uncertainty Considerations in WLS State Estimation
,
2009,
IEEE Transactions on Power Systems.
[4]
R. Jabr,et al.
Statistical Representation of Distribution System Loads Using Gaussian Mixture Model
,
2010
.
[5]
E.I. Amoiralis,et al.
Transformer Design and Optimization: A Literature Survey
,
2009,
IEEE Transactions on Power Delivery.