No abstract

No abstract

We develop data-driven dynamical models of the nonlinear aeroelastic effects on a long-span suspension bridge from sparse, noisy sensor measurements which monitor the bridge. Using the sparse identification of nonlinear dynamics (SINDy) algorithm, we are able to identify parsimonious, time-varying dynamical systems that capture vortex-induced vibration (VIV) events in the bridge. Thus we are able to posit new, data-driven models highlighting the aeroelastic interaction of the bridge structure with VIV events. The bridge dynamics are shown to have distinct, time-dependent modes of behavior, thus requiring parametric models to account for the diversity of dynamics. Our method generates hitherto unknown bridge-wind interaction models that go beyond current theoretical and computational descriptions. Our proposed method for real-time monitoring and model discovery allow us to move our model predictions beyond lab theory to practical engineering design, which has the potential to assess bad engineering configurations that are susceptible to deleterious bridge-wind interactions. With the rise of real-time sensor networks on major bridges, our model discovery methods can enhance an engineers ability to assess the nonlinear aeroelastic interactions of the bridge with its wind environment.

Summary A data‐driven approach for modeling bridge buffeting in the time domain is proposed based on the structural health monitoring (SHM) system. The long short‐term memory (LSTM) network is applied to model the bridge aerodynamic system with the potential fluid memory effect which is characterized by an uncertain time lag between inflow wind and the structural response. SHM is incorporated into this data‐driven approach due to the advantages of prototype measurements such as the ability to consider the high Reynolds number effects and the real natural winds with nonuniformity and nonstationarity. The cell state in the LSTM module is applied to carry the potential fluid memory effects for predicting the aerodynamic response. We compare the obtained data‐driven model and the conventional finite element model in the buffeting response prediction. The data‐driven model shows higher accuracy than the conventional model, indicating that the proposed data‐driven approach has promising potential in modeling bridge aerodynamics. The incorporation of the proposed LSTM‐based bridge aerodynamic model and the field monitoring enables us to move buffeting predictions from lab theory to practical engineering.

Vortex-induced vibrations (VIVs) with large amplitudes have been observed on long-span bridges worldwide. Classic semi-empirical VIV models that depend on wind tunnel tests are challenged when required to predict the VIV response of real bridges due to the complexity of real winds, high Reynolds number effects, and uncertainty of bridge structures. The prediction accuracy by these laboratory-based models may, thus, be reduced for real large-scale bridges. Emerging field monitoring systems on prototype bridges allow one to reconsider modeling of bridge VIVs with considerations of real natural winds and full-scale structures by massive monitoring data. In this research, first, we derive a general form of time-dependent ordinary differential equation based on Scanlan's semi-empirical model and field observed bridge VIVs to describe VIV dynamics. Second, guided by the formulation and field observation, we propose a deep learning framework to identify the VIV dynamics, leading to a data-driven model. We demonstrate the proposed framework on a real long-span bridge by performing long-time prediction of the VIV response under real natural winds.