Emerging Technologies in Power Systems: Natural Oscillation Cancellation and Sinusoidal Waveform Processing

In recent years, power systems have been undergoing notable changes, including the rapid growth of renewable energy and inverter-based resources (IBRs), and the interconnection of large loads, e.g. data centers. These changes answer the growing and changing requirements of human society and industry but also bring new challenges to power systems. This dissertation investigates a few of these challenges and proposes solutions.

To help deal with the increasing amount of switching events in power systems and the decreasing amount of physical inertia, Chapter 2 proposes a novel countermeasure for natural oscillations, oscillation cancellation (OC), which is different from the traditional damping approaches. It can neutralize natural oscillations induced by switching operations at the point of interconnection and prevent them from entering the power system. Chapter 3 first provides a few different perspectives to explain and visualize the underlying mechanism of OC. Then, three enhancement techniques: mode weighting factors, predictive OC, and distributed OC, are proposed to further improve the performance of OC proposed in Chapter 2. To help improve and benchmark IBR modeling, Chapter 4 proposes an IBR electromagnetic transient (EMT) model verification solution using playback simulation, and two key techniques that facilitate this solution. To tackle the increasingly challenging anomaly detection task as power systems become larger and more complex, Chapter 5 proposes a generic approach for sinusoidal waveform anomaly detection and visualization. It offers finer scale monitoring and an adjustable sensitivity, while only inducing minimum computation. To streamline and automate outage event response processes and help reduce power restoration time after outages, Chapter 6 proposes a set of waveform pre-processing techniques to prepare waveform data for being used as inputs to machine learning algorithms, and a machine learning algorithm based on neural networks to classify outage waveforms according to their root causes. Finally, Chapter 7 concludes this dissertation and discusses potential topics for future work.