A Decomposition Method for the Total Leakage Current of MOA Based on Multiple Linear Regression.

A new current decomposition method based on multiple linear regression is presented. The time-domain equations of every current component on the applied voltage are deduced based on an improved equivalent model of the metal–oxide surge arrester (MOA). The capacitive current compensation method and the proposed method are used to decompose the total leakage current and to study the influence of the nonlinearity of MOA model, third harmonic voltage, and white noise on the decomposition accuracy. The simulation results show that the proposed method can exactly decompose the resistive current and the capacitive current if the applied voltage and the total leakage current are noiseless. The proposed method is more accurate than the capacitive current compensation method. The total leakage current of an MOA varistor is measured and is then decomposed by both methods. The experimental results verify the effectiveness of the proposed method. [ABSTRACT FROM PUBLISHER]