Fault Location Algorithm for Non-Homogeneous Transmission Lines Considering Line Asymmetry

This article presents a novel phasor domain approach to determine the location of faults in non-homogeneous transmission lines. Best implementation requires synchronized voltage and current phasor measurements at all terminals of the transmission line. The overall non-homogeneous transmission line model with fault is established by systematically combining the generalized compact models of all homogeneous line sections. The location of the fault is introduced as an additional state of the overall model. The derivation procedure of the model is without further assumptions, resulting in full consideration of three phase line asymmetry as well as distributed parameters of non-homogeneous transmission lines. Afterwards, the location of the fault is identified using the state estimation algorithm. Extensive numerical experiments in a two-terminal and a multi-terminal non-homogeneous transmission line demonstrate that the method has higher accuracy than existing fault location methods, independent of fault types, locations and impedances.

I. Introduction

Accurate FAULT location on transmission lines benefits operators and utility crews by minimizing labor, repairing costs and power outage time [1]–[3]. In recent years, a number of two-terminal and multi-terminal non-homogeneous (also known as compound, mixed, or hybrid) transmission lines have emerged in the modern power systems, where the overhead transmission lines are combined with underground cables, to connect off-shore wind farms to the existing grids, mitigate right-of-way related issues, etc. Nevertheless, accurate fault location in non-homogeneous transmission lines is challenging due to the topology complexity, and has been studied by a limited number of literatures. Existing fault location techniques can be mainly categorized into two groups: fundamental frequency phasor based methods and traveling wave based methods. Besides the two main groups of methods, artificial intelligence based methods are also studied in recent years [19], [30]. These methods usually require a large number of high-quality training data as well as heavy training burden and therefore have not been widely adopted in practice. Next, the two main groups of methods are reviewed for classic transmission lines with commentary on specific challenges for non-homogeneous transmission lines.