Fast fault detection scheme for series-compensated lines during power swing

This paper reports the use of a novel ultra-high speed scheme to release the distance relay to operate for a fault during a power swing in the series compensated line. In the scheme, in order to extract the fault induced voltage and current components, voltage and current samples are analyzed by the multi-resolution morphological gradient (MMG), first. Then, the fault initiated forward travelling wave is computed at the distance relay point. Next, Likelihood ratio [LR] test is utilized to detect a jump in the statistical mean of the calculated forward travelling wave. Finally, a support vector machine (SVM) classifier is employed to distinguish faults from other normal capacitor and switching transients. It is shown that in all of the simulated cases, our ultra-high speed algorithm was successful in fault detection across a wide range conditions including, fault type, fault resistance, fault location, pre-fault loading and fault inception time. Moreover, we found that using the proposed scheme significantly speeded the fault detection, in comparison with the existing phasor based methods. In addition, the improvements noted in our algorithm are achieved with a low computational burden.