Evaluation method of nonlinear customer harmonic emission level based on correlation vector machine

Abstract

A high cable ratio of urban distribution network is easy to cause the problem of harmonic resonance. Nonlinear customers inject harmonics into the power grid in operation, resulting in harmonic issues. This study proposes a method of harmonic emission level based on correlation vector machine regression. Firstly, the Norton equivalent circuit of customer is established. The harmonic impedance at the system side is obtained via the correlation vector machine regression model utilizing the metrical data of the common connection point as the input vector, and the harmonic emission level of nonlinear customers is then estimated. The proposed method overcomes the problems of support vector machine method, kernel function must meet Mercer condition, and parameters are difficult to determine. The effectiveness of the proposed method is verified by simulation analysis.