Quantitative comparison of partial discharge localization algorithms using time difference of arrival measurement in substation

Partial discharge (PD) non-contact detection by using antenna array has gained extensive attention in air-insulated substation (AIS) monitoring. One significant superiority is that the positions of PD sources can be located by utilizing the differences of received signals. Also locating electromagnetic/acoustic radiative sources using passive sensor arrays is a continuous theme in the field of radar and sonar, and various localization methods have been developed. In this paper, we mainly concern the time difference localization techniques, which can be classified as maximum-likelihood (ML) and least squares (LS) estimators. The ML estimates positions by maximizing the probability density function, which is a high-nonlinear problem and difficult to be solved. The iteration or simplification solutions are proposed to give approximate ML results. As an alternative, LS estimators which can give closed-form solutions and with high computation efficiency are widely adopted. In this research, the theories and calculation procedures of several kinds of localization algorithms are reviewed, i.e., approximate ML, probability-based method, spherical-interpolation, Chan and squared range-difference LS, and their performances are compared by Monte Carlo simulation and experiments. The results indicate that probability-based method presents highest accuracy and computation efficiency. Among other methods, Chan, bias-reduction and squared range-difference provide satisfying localization performance.