Accurate Estimation of Partial Discharge Location using Maximum Likelihood

In this article, a novel algorithm for high-accuracy partial discharge (PD) localization in an oil insulation system is proposed. This study aims to identify the statistics of the dynamics and measurement noise sequences in a PD localization system and employ that information in realizing better estimates. An extended Kalman filter (EKF) is used to estimate the PD location. The performance of the filter is enhanced by identifying the true statistics of the noise sequences using a maximum likelihood estimation approach. The accuracy of the proposed optimal algorithm is verified experimentally by estimating the PD location in an oil insulation system under two experimental settings.