No abstract

This paper presents a new technique for accurate fault locator based on synchronized voltage measurement and smooth support vector machines (SSVM) HV teed feeder transmission line. The approach consists of detection of faulted branch, classification of fault type and determination of exact fault location. Post-fault measured voltages waveforms are collected from only two ends of the three branches teed feeder system. The application of SSVM (Classification and Regression) is practiced for training, testing and validating of the faulted waveforms data set leading to the exact fault location on the system. Several fault conditions are analyzed, trained, tested and validated. The proposed technique is tested and found insensitive to variation of different parameters such as fault type, fault resistance and fault inception angle. ATP-EMTP program is used for simulation of faulted data for a 275KV teed feeder transmission system.

No abstract