Generalized Regression Neural Networks Based HVDC Transmission Line Fault Localization

Abstract

In this paper, a line fault location algorithm based on singular value decomposition and generalized regression neural networks (GRNN) is proposed for high voltage direct current (HVDC) transmission. As we know, the arriving instants of the first fault-induced transient backward travelling wave and the reflected wave can be detected. The high dimensional feature of travelling wave is different conditional on the term of diverse line fault position. Therefore, the fault distance can be estimated by using the velocity or current of the travelling wave. Firstly, we use singular value decomposition (SVD) method to extract the feature of HDVC travelling wave. After that, the features are sent to GRNN to model the relationship between the travelling wave and line fault position. For the sake of simplicity, the proposed algorithm is shorted for SVD-GRNN in the rest of this paper. Finally, simulation result indicates that line fault position can be accurately localized by the proposed algorithm.