The Application of Levenberg-Marquardt Algorithm in On-Line Monitoring for XLPE Cables

Abstract

The insulation condition of XLPE cables could be evaluated comprehensively by its working insulation resistance, equivalent capacitance and dielectric loss factor. To calculate the three parameters precisely, an improved data processing method was presented, in which the nonlinear least-squares fitting algorithm was utilized based on Levenberg-Marquardt algorithm that took the results of fast Fourier transform as the initial value for iteration. The parameters of fundamental component were extracted from the sampled signals quickly, and the influences caused by the frequency fluctuation, harmonic content and direct current content were restrained well using this method. The simulation results were given to show the effectiveness of this method. The reliability of on-line monitoring could be promoted.