Ultrasonic partial discharge detection using negentropy-based matched filter

It is essential to detect and localize partial discharge in a nondestructive test, which ensures reliable, long-term operation of critical power apparatus under high voltage stress. In this paper, the ultrasonic detection method combined with electrical PD detection method is used to detect the typical internal PD in transformer. The ultrasonic signals are polluted by various noises when power apparatus operate. The matched filter based on negentropy is presented to detect the ultrasonic PD signal hidden in noises. The simulation and experimental tests demonstrate that the matched filter proposed in the paper suppress the interference signals effectively, and extract the clearly ultrasonic PD signal to enhance the reliability of identification. By producing three-dimensional distributions, it is proved the validity of the ultrasonic detection method.