An Intelligent Detection of High-Impedance Faults for Distribution Lines Integrated With Distributed Generators

The high-impedance faults (HIF) occurring on distribution lines go undetected by the conventional relays due to their low current magnitude. In addition to low current, HIFs also exhibit typical characteristic features, such as randomness, nonlinearity, etc., which are similar to the characteristics of the switching events, such as harmonic load switching. The advent of distributed generators (DG) into the distribution system, changes the current contribution through the substation, which also affects the HIF detection. This paper presents an intelligent HIF detection technique for distribution lines incorporating the DGs. Variational mode decomposition (VMD) is used to extract the input patterns. The commonly used empirical mode decomposition has limitation of modal aliasing that is overcome by VMD. VMD decomposes the signal into different modes that are further converted to singular values so as to obtain the analytic signal. These are finally used as an input to an efficient classifier known as the support vector machine, to create an intelligent classifier. The proposed methodology is verified on a modified IEEE 13-bus feeder for various events considering the changes in DG parameters and also the noise levels.