合格电网节点的特征化用于电力系统振荡识别

Rapid integration of renewable energy sources and power electronic components in the conventional energy systems has resulted in increased sub-synchronous oscillations (SSOs) in the power network. As a consequence, accurate and exhaustive monitoring of SSOs is critical for reliable system operation. The importance of failure-proof network operation has necessitated the requirement of a revised notion of power system stability and control. In this regard, this paper presents a benchmark study on power grid disturbances to efficiently identify the most qualified buses that should be monitored to capture the perturbations present in the network. Our analysis shows that the proposed approach is computationally much faster and utilizes a significantly reduced number of buses for disturbance identification as compared to the state-of-the-art, without compromising on the disturbance identification accuracy in the network. Theoretical formulation of optimization objectives are verified by the results from the data-driven performance results using phasor measurement units (PMUs). PMU data was generated through real-time structured computer-aided design (RSCAD) simulation. Considering various adversities, such as single PMU loss, single line loss, measurement noise-infested PMU data, and compromised PMU operation, our results demonstrate 50% and 75% reduced data footprint under normal and adverse system operations, respectively, while identifying > 95% of the critical SSO frequencies, with a >94% reduced execution time. Also, the proposed approach presents 25% enhanced accuracy of SSO identification in standard IEEE networks, 25% improved noise immunity, 15% additional immunity to grid adversities as compared to the state-of-the-art machine learning and statistical SSO identifiers, with only 0.21% accuracy loss on per-node increase in grid size.