Detection and location of partial discharges in power transformers using acoustic and electromagnetic signals

TLDR

Partial discharge measurement is a non‑destructive diagnostic tool that enables detection and localization of faults, and locating the PD source is critical for efficient maintenance and risk assessment. The study proposes a new acoustic‑based localization method that uses pseudo‑arrival times and robust direct solvers to geometrically locate partial discharges. The method relies on precise acoustic arrival time measurement, automatic objective time extraction, wavelet‑based denoising, and pseudo‑time calculations to improve location accuracy. Averaging acoustic PD signals increases sensitivity, lowers the detection limit, and enables arrival time determination for weaker discharges.

Abstract

The measurement of partial discharges (PD) is a non-destructive and sensitive diagnostic tool for the condition assessment of insulating systems. Two major tasks of PD measurements may be distinguished, (i) PD detection, hence providing evidence and the type of the PD and (ii) the location of the PD. The question "Where is the PD source?" is amongst others tremendously important for scheduling and starting maintenance/repair actions cost and time efficiently or to perform a risk analysis. Here the possibility to geometrically localize the flaw, by means of arrival times of acoustic PD signals, gets an extremely interesting option. Precise acoustic arrival times are consequently essential to accurately locate PD in a power transformer. The averaging of acoustic PD signals helps to enhance the acoustic sensitivity. The acoustic detection limit is lowered significantly and the determination of the arrival times is made possible for weaker PD. Supplementary steps, like automatic objective arrival time determination or additional wavelet-based de-noising further improve the overall location accuracy. A new location approach works with pseudo-times and allows for the use of robust direct solvers instead of the previously used iterative algorithms.