With the development and installation of Phasor Measurement Unit (PMU), time-synchronized high-resolution data is available at the control center for enhanced visibility of the power system transmission network.

North American Electric Reliability Corporation (NERC) identifies malfunctions of protection devices as one of the top concern for the large power grid blackout events. A failure diagnosis method is needed which can hypothesize the malfunction of devices and find the root cause of this malfunction. Also, some recent research work shows that protection assets could suffer from glitches or downgrade over time leading to a device failure if not addressed timely. It is important to detect such downgrades/glitches at the early stage to help the maintenance and repair work.

In this dissertation, a multi-hypothesis theory based failure diagnosis algorithm is developed for transmission network protection system. The proposed centralized failure diagnosis tool (FDProSys) analyzes the synchrophasor data to provide multiple hypotheses to explain the observed ambiguous condition in the transmission network protection system. And the hypothesis, which is the most likely correct will be selected for the root cause analysis. The developed approach can do failure diagnosis and root cause analysis automatically and efficiently.

The developed tool (FDProSys) can also detect anomalies in synchrophasor data to start the failure diagnosis method in physical part and apply similarities based substation level cyber intrusion detection for cyber part simultaneously. The results from both cyber and physical part are used to conclude the root cause of the observed anomalies.

Additionally, a clustering method based protection asset monitoring technique is developed in this work. The developed monitoring method utilizes the real-time synchrophasor data collected from the field and monitors the health of the protection asset by detecting cluster changes in the streaming data. The tool employs topology information and circuit breaker status to differentiate system events/faults from asset downgrades.

The effectiveness of the proposed techniques is validated with different transmission system with distance protection models, e.g., IEEE 14-bus system, 57-bus system, and 300-bus system and found to be satisfactory.