Partial discharge localization in transformers using UHF sensors

The location of the partial discharge (PD) source inside a transformer can be determined by using the time differences between signals that are captured by an array of three UHF sensors. However, other unwanted signals are also picked up by the sensors and may result in a false PD location. This paper discusses the application of multivariate denoising to the PD signals. Three methods are compared to determine the time difference between the received signals and locate the PD source. With the first peak method, the time difference is calculated by taking the first peak of the signal as the arrival instant. In the second method, the arrival time is determined from the knee point of the cumulative energy curve. For the third method, the time difference is determined by applying the similarity function to the plot of the cumulative energy of the PD signals. The last method is fully automated. However, experimental results show the first peak method tends to be the most accurate.