Travelling Wave Fault Location Based on Pattern Recognition: Application to Complex Topologies and Field Records

Type-D travelling wave fault locators have multiple sources of error, most of which depend on the knowledge of the line parameters and on the measurement hardware. Two of those sources of error can be reduced with a better algorithm: the error caused by non-homogeneous lines, and the error on the detection of the travelling waves arrival times. In this paper, an updated type-D algorithm and a pattern recognition algorithm are presented. They are applied to simulation models that include attenuation and distortion, reflection on elements outside of the line, T-junctions, and non-homogeneous lines to illustrate the different sources of error. In addition, the algorithms are also applied to a fault record acquired in the Belgian transmission network.