Original title: PINNs-Based Uncertainty Quantification for Transient Stability Analysis
Authors: Ren Wang, Ming Zhong, Kaidi Xu, Lola Giráldez Sánchez-Cortés, Ignacio de Cominges Guerra
This study tackles power system stability during transients, where parameters are uncertain. It introduces a creative use of Physics-Informed Neural Networks (PINNs), called Ensemble of PINNs (E-PINNs), to precisely estimate vital factors like rotor angle and inertia amidst uncertainties. Leveraging the physics-based swing equations, E-PINNs offer accurate parameter estimations and provide insights into potential uncertainties, enhancing our understanding of system behavior. Demonstrated on $1$-bus and $2$-bus systems, E-PINNs excel in handling parameter variability and limited data. This pioneering approach in machine learning for power systems advances transient stability analysis, offering more dependable and computationally efficient methods.
Original article: https://arxiv.org/abs/2311.12947