Modeling the Inverse Problem of Dielectric Response of Oil-impregnated-paper Insulation

Abstract

As a non-intrusive measurement method, frequency domain spectroscopy (FDS) is widely used to diagnose the global defects (such as aging and dampness) of the insulation body in power equipment. However, the FDS method can only map the holistic dielectric response (DR) characteristics of composite oil-impregnated-paper (OIP) insulation that usually comprises both the insulation paper and oil. Under irregular insulation structure, the DR of each dielectric cannot be simply distinguished, which is unconducive to detect the insulation weakness therein. In this paper, a novel inverse problem model to analyze the separate DR of each dielectric within a compound insulation structure is proposed. Built on the modified Cole-Cole formula, the inverse problem is solved by an optimization process which combines the finite element method and GA-LMA numerical algorithm, and the dielectric variables of insulating oil and paper are back-calculated respectively. By comparing the error of the measured and reconstructed FDS traces, it is evidenced that the proposed model of DR inverse problem and its solution method are suitable for FDS measurements, which signifies new indications for refined state evaluation of OIP insulation by enabling inspecting the separate DR information of each dielectric within a real-life complex insulation structure.