Automated Identification of Electrical Disturbance Waveforms Within an Operational Smart Power Grid

Electric power utilities employ “smart” re-closing devices capable of capturing electrical disturbance (ED) waveforms in real-time. These waveforms are digitally sampled, transmitted back to the utility, and stored in text files that comply with the IEEE COMmon format for TRAnsient Data Exchange (COMTRADE) standard. An average of 2,100 ED events are recorded, by 850 devices, on a monthly basis within Electric Power Board's distribution network. Due to operational restraints such as: time, money, and person power, only 2% of these recorded events are currently analyzed by power utility personnel. The remaining 98% of recorded events remain a significant loss in information, which could be vital in fully understanding the operating condition within the distribution network. This work presents a hierarchical automated classification process that assigns ED events into one of three “categories”: valid data, switching events, and faults/power quality (PQ) disturbances. This process correctly assigned 92.9% of the tested 140 COMTRADE files into one of these categories. In addition to this categorization, this work presents an automated approach for the classification of magnetic inrush events. A classification rate of 90.63% is demonstrated using 32 transformer inrush COMTRADE files.