Online Source Location of Forced Oscillations in Power Systems Using Synchrophasor Measurements

Bikal Pudasaini

doi:10.7273/000006361

The electricity grid stands as a remarkable accomplishment in modern engineering, facilitating the transmission of power across extensive distances from diverse generation sources to end consumers. It serves a crucial role in supplying electricity to an array of establishments, including residential, commercial, and institutional buildings, ensuring uninterrupted access to power throughout the year, day and night. However, this intricate system also possesses inherent complexities and susceptibility to stability problems that may result in system breakdowns and widespread power outages. This dissertation focuses on a specific stability concern: Forced Oscillation. The location and mitigation of low-frequency forced oscillation are important for the reliable operation of the power system. These are sustained oscillations caused by external periodic input and may resonate with poorly damped natural modes across a wide region of the power system. As a result, they pose a considerable risk to system reliability, potentially leading to system failures and subsequent power outages. Accurate source location is a crucial prerequisite for the successful implementation of mitigation schemes to address forced oscillation. This dissertation presents two distinct and effective methods for accurately locating the source of forced oscillation using PMU measurements. The first method is an automatic approach that analyzes the initial transient at the onset of a forced oscillation event. It exploits the concept that signals at or near the source of the FO will show faster changes in the signal responses at the start of the oscillation, in contrast to signals that are farther away from the source. The second method introduces an unsupervised machine learning framework based on the isolation forest algorithm, utilizing PMU measurements. It formulates the problem as an anomaly detection task, with the objective of locating the source of forced oscillation as an anomaly. Feature extraction is performed using the Hankel least squares method. This dissertation provides a thorough description of both methods, encompassing their motivations, methodologies, core concepts, algorithms, and mathematical equations. Comprehensive step-by-step implementation frameworks are presented, along with the necessary parameters. Additionally, guidance on parameter tuning is provided. To evaluate the effectiveness of the methods, various test cases are conducted on the 240-bus WECC test case system. The results obtained from the methods are compared with state-of-the-art and widely-used approaches in the existing literature to showcase their robustness. The dissertation also acknowledges the limitations and assumptions underlying the methods, ensuring the correct implementation of the proposed approaches. Furthermore, future directions for research and development in this area are outlined, providing a roadmap for further advancements.

Online Source Location of Forced Oscillations in Power Systems Using Synchrophasor Measurements

Files and links (1)

Abstract

Metrics

Details