Murat Furkan Mansur, Tufan Kumbasar · Apr 16, 2026
SOLIS is a physics-informed learning approach for nonlinear system identification, combining the interpretability of classical methods with the expressiveness of neural networks. It models unknown dynamics using a state-conditioned second-order surrogate model and recasts identification as learning a quasi-linear parameter-varying representation. SOLIS achieves accurate parameter-manifold recovery and coherent physical rollouts from sparse data, outperforming standard inverse methods in challenging regimes.
Why This Matters
This paper matters for power system engineers as it proposes a novel approach to nonlinear system identification that can recover interpretable parameters without presupposing a global equation, which is particularly relevant for tackling complex power system dynamics and improving grid resilience. The proposed SOLIS method can be applied to various grid operations tasks, such as ISO operations, FERC filings, and NERC standards compliance, by providing more accurate parameter estimation and physical rollouts from sparse data.