Data-driven analytics for real-time monitoring of power systems using synchrophasor technology

Abstract

Modernising power grid monitoring through Phasor Measurement Units has unlocked a new paradigm for situational awareness. The synchrophasor measurements obtained from Phasor Measurement Units usually reporting at 25/50 or 30/60 samples per second for a 50/60 Hz system, provide a high-resolution and time-aligned overview of the system’s dynamics. Data from synchrophasor measurement devices is large, therefore creating operational challenges due to the limited fulfillment of real-time analytics and monitoring applications at a control center, which are essential for enhanced situational awareness. Events lead to blackouts, therefore timely identifying events and assessing the system’s stability becomes essential for faster system restoration. Recent trends have shown the importance of machine learning based techniques to offer a good generalising capability for different applications of power systems. Therefore, the thesis explores on developing machine learning based applications for classifying transmission line faults and also develop machine learning models that can assess the transient stability status post transmission line faults. Assessing the impact of grid events on transient stability becomes even more critical due to the increasing penetration of renewables, which is responsible for diminishing the stability margins, which may further aggravate the impact of events on the power systems.