Quantitative Comparison of Partial Discharge Localization Algorirthms on Power Transformers Based on Acoustic Method

The localization of partial discharge (PD) sources using the acoustic emission(AE) technique has attracted increasing research attention. Previous studies have focused on applying different optimization algorithms to solve the time difference of arrival equations. Some simple iteration algorithms and some intelligent optimization algorithms are used to solve the time difference of arrival equations. Moreover, a new localization method using transformer model is proposed. In this paper, experiments are performed in a 35kV transformer to compare quantitatively those different localization methods. Five different localization methods are introduced in detail about their algorithm theory, advantages and disadvantages. After that, those algorithms are tested using data from a three-phase 35kV transformer. To generate predictable PD signals, an artificial protrusion defect model was used. This defect was placed at different positions in the transformer and the localization results of various algorithms can be gathered. Both the Newton-Raphson algorithm and the Chan algorithm suffer non-convergence or localization outside the transformer. The GA and ICA have better performance but large errors are induced because those two methods treat direct paths as the fastest routes. Moreover, the algorithm based on the transformer model shows the best accuracy but its calculation time is the longest while other methods can response quickly.