Estimation of Power-Coefficient Curve from SCADA Data for Digital-Twin Applications

Digital twins are emerging as a pivotal technology for the performance optimization, predictive maintenance, and real-time monitoring of wind turbines. However, the accuracy of these virtual representations critically depends on the availability of the power coefficient ($C_p$) curve, a key descriptor of a turbine’s aerodynamic efficiency. This information is often proprietary and not disclosed by manufacturers, posing a significant barrier to the development of high-fidelity digital twins. This study addresses this critical gap by proposing a novel framework for estimate $C_p$ curves using operational Supervisory Control and Data Acquisition (SCADA) data. The proposed methodology utilizes a parameterized mathematical formulation to model the $C_p$ curve and employs the Adam optimizer to robustly tune the model’s parameters against real-world operational data. The framework was evaluated through a two-pronged process. First, the model’s accuracy was assessed using synthetic SCADA data from a high-fidelity simulator under ideal conditions, demonstrating excellent agreement with an $R^2$ exceeding 0.99 and a normalized Mean Absolute Percentage Error (nMAPE) ranging from 4.38% to 6.03%. Second, its practical performance was evaluated using real SCADA data from a commercial wind turbine, where it maintained high accuracy with an $R^2$ ranging from 0.89 to 0.98 and an nMAPE of 3.27% to 5.97%. The findings demonstrate that the proposed methodology can effectively reconstruct a turbine’s aerodynamic characteristics without proprietary manufacturer data. This research offers a viable pathway for operators and researchers to create accurate, turbine-specific digital twins, thereby enabling enhanced performance monitoring, advanced control optimization, and predictive maintenance for more efficient and reliable wind energy production.