WorksEngineering
Compositional Data-Driven Small-Signal Stability Assessment of Power Systems via Output Differential Passivity
Abstract
This paper proposes a compositional data-driven small-signal stability assessment framework for interconnected power systems based on output differential passivity (ODP). Without requiring explicit device models, local ODP matrices and scalar indices are estimated directly from measurement trajectories. When internal states are available, trajectory-based linear matrix inequalities (LMIs) are constructed from input-state measurements to estimate the local ODP matrices and scalar indices. When internal states are unavailable, an extended-state representation is constructed from input-output measurements, enabling the same assessment using only external signals. The estimated ODP matrices are combined with the network coupling matrix for compositional assessment of power system stability, while their scalar indices enable decentralized assessment. To address the computational and numerical challenges of trajectory-based LMIs, we introduce right-inverse whitening (RIW) to remove redundant behavioral coordinates and improve computational efficiency, and apply instrumental-variable (IV) preprocessing to alleviate numerical ill-conditioning and mitigate noise effects. Case studies on a modified IEEE 39-bus power system show that ambient measurements satisfy the required informativity conditions and that both formulations yield estimates close to model-based references under the tested noise levels. The reusable local ODP estimates also enable modular stability reassessment after controller updates.