Data-driven Aerodynamic Analysis of Structures using Gaussian Processes

An abundant amount of data gathered during wind tunnel testing and health monitoring of structures inspires the use of machine learning methods to replicate the wind forces. These forces are critical for both the design and life-cycle assessment of lifeline structures such as bridges. This paper presents a data-driven Gaussian Process-Nonlinear Finite Impulse Response (GP-NFIR) model of the nonlinear self-excited forces acting on bridges. Constructed in a nondimensional form, the model takes the effective wind angle of attack as lagged exogenous input and outputs a probability distribution of the aerodynamic forces. The nonlinear latent function, mapping the input to the output, is modeled by a GP regression. Consequently, the model is nonparametric, and as such, it avoids setting up the latent function's structure a priori. The training input is designed as band-limited random harmonic motion that consists of vertical and rotational displacements. Once trained, the model can predict the aerodynamic forces for both prescribed input motion and coupled aeroelastic analysis. The presented concept is first verified for a flat plate's analytical, linear solution by predicting the self-excited forces and flutter velocity. Finally, the framework is applied to a streamlined and bluff bridge deck based on Computational Fluid Dynamics (CFD) data. Here, the model's ability to predict nonlinear aerodynamic forces, critical flutter limit, and post-flutter behavior are highlighted. Further applications of the presented framework are foreseen in the design and online real-time monitoring of slender line-like structures.