Author(s): Weichao Ma; Wei Zeng; Xu Lai; Jiandong Yang

Linked Author(s):

Keywords: Hydropower; Stability analysis; Physics-informed neural network; System identification; Model calibration

Abstract: Ensuring the stability of hydropower generating systems (HGSs) is critical for their flexible operation in modern energy systems. However, accurately predicting the dynamic processes of HGSs remains a significant challenge due to the simplifications inherent in physics-based models. To address this issue, this study proposed a physics-based and data-driven method to reconstruct hydraulic parameter variations during the dynamic processes of HGSs. The proposed method integrates a physics-informed neural network (PINN) that combines observed data with the physical functions of HGSs. This enables the PINN to learn and extract hidden information embedded in the observed data. Once trained, the PINN can predict the complete dynamic characteristics of HGSs, even when simplified physical functions are used. A numerical case study demonstrates the effectiveness and advantages of this approach.

DOI: https://doi.org/10.64697/978-90-835589-7-4_41WC-P2098-cd

Year: 2025