Three-dimensional Reconstruction of the Electric Tree in Oil-impregnated Pressboard Based on Partial Discharge Location

摘要

There is still a lack of effective means of observation for electrical trees in opaque materials, which presents a great obstacle to the accurate diagnosis and prediction of partial discharge (PD) defects. As an effective tool to solve the field source, inverse problem has also been applied to the reconstruction of electrical tree. Therefore, the accuracy of solving the inverse problem of electrostatic field is verified theoretically. Firstly, the PD is equivalent to point charge, and the inverse problem equation is established, which takes the position of the point charge as the unknown quantity and the field strength of the measured point as the known quantity, and is solved by PSO. The error range of PSO can be calculated by randomly changing the position of charge source, and then 10 groups of noises (1% ~ 10%) can be added to the electric field value of the measurement point to fix the position of charge source. The influence of different noise levels on the calculation error and the variation rule of positioning error with noise level are analyzed. Through statistical analysis, it is concluded that the error range of each variable calculated by PSO is very small $(< mathbf{4.5}%)$, which verifies the accuracy of inverse problem solving; the calculation error of variables $pmb{x},pmb{y},pmb{z}, pmb{q}_{mathit{1}}$ increases gradually with the increase of noise level, and the root mean square error of each coordinate $pmb{S}_{pmb{x}} > pmb{S}_{pmb{y}} > pmb{S}_{pmb{z}}$.