Root Cause Identification of Power System Faults using Waveform Analytics

A methodology intended to automatically identify the underlying cause of fault in transmission lines is presented. Fault data recorded by different Intelligent Electronic Devices (IEDs) in Common format for Transient Data Exchange for power systems (COMTRADE) is replayed to extract statistically significant features. Time domain features of the recorded data have been investigated. A fault triggering algorithm based on voltage sag/swell is used for segmentation of pre-fault, fault and post-fault period in the recorded data. One-way Analysis of Variance (ANOVA) is used to find the required confidence interval to validate the statistical significance of the selected features. The proposed method uses machine learning approach to classify faults, as they occur in the system, into preselected fault root cause groups. The developed algorithm has been tested using real field dataset. The results show that the methodology is sound and covers a wide range of root causes of faults.