Detecting the location and severity of transformer winding deformation by a novel adaptive particle swarm optimization algorithm

Abstract

Transformers as 1 of the most important and expensive equipment of power network are under continuous exposure of faults. Periodic examinations for detecting and locating trivial deformation occurrences are indispensable for preventing their unexpected outage and costly, time-consuming repair procedure. The most crucial step for fault detection is obtaining parameters of the detailed model. Acquiring parameters of the detailed model (APDM) is a nonlinear, nonconvex, and large-scale problem that requires a powerful method to be solved. This paper proposed a novel method that gains benefit from new efficient functions of Fourier series coefficient and phase functions, an adaptive version of the PSO and fast calculation of coupled network matrices, to solve the problem. Experimental results on 2.64 and 120 kV set of transformer windings, constructed for this study, verified the precision of the proposed method for discovering the position and severity of the disk space variation fault.