Channel: Special Column on Applications of Artificial Intelligence in Bridge Wind Engineering http://zgglxb.chd.edu.cn EN-US http://zgglxb.chd.edu.cn/EN/current.shtml http://zgglxb.chd.edu.cn 5 U_j, moving velocity U_T, and ambient wind speed U_b are crucial to the time-varying mean wind speed of the downburst. The parameter optimization efficiency significantly improves with the introduction of the surrogate model, which participates directly in the iteration of optimization instead of the numerical model. The numerical simulation of downbursts based on the optimized parameters notably improves the accuracy.]]> n moments under different wind attack angles were the input and the aerodynamic coefficients of the n+1 moments were the output. First, an LSTM network framework was constructed based on the open-source TensorFlow library. Second, training and test sets were constructed based on the numerical simulation results of the unsteady aerodynamic forces on a bridge at three angles of attack (3°, 4°, and 5°), and the models were trained. Finally, in the interpolation prediction, a dataset was constructed from the aerodynamic coefficients at angles of attack of 3° and 5° to train the model, and the aerodynamic coefficient of 4° was predicted. For extrapolation prediction, a dataset was constructed from the aerodynamic coefficients at angles of attack of 3° and 4° to train the model, and the aerodynamic coefficient of 5° was predicted. The results show that the LSTM prediction results are basically consistent with the trajectory, amplitude, and period of the numerical simulation results over time. To verify the generalization of the model further, the trained LSTM extrapolation prediction model was used to predict the aerodynamic coefficients at angles of attack of 0° and 3°. The results show that the model can predict the original results better. Therefore, this algorithm has potential applications in predicting the unsteady aerodynamics of bridge structures and can provide technical support for engineering applications.]]> ® software with mesh deformation drive technology were used to obtain an initial set of representative points in the design space. The training data pool was obtained via a computational fluid dynamics (CFD)-based numerical simulation. Then, a WOA-GRNN surrogate model related to the design parameters of the wind fairing and vortex vibration response was developed and trained through the obtained CFD data pool. Next, a typical long-span bridge with Π-shaped girder was used as the engineering background to predict the VIV response surface of the bridge in the multi-parameter design space using the trained WOA-GRNN surrogate model. Finally, the correctness of the surrogate model was verified by comparison with wind tunnel test results. The results show that the WOA-GRNN surrogate model can significantly improve the efficiency of bridge aerodynamic shape optimization. For the Π-shaped beam in the studied case, the vertical VIV response of the optimal wind fairing aerodynamic shape #1 is 99% lower than that of the prototype section; in addition, the VIV suppression effect is excellent. This study can provide important references for future VIV control, aerodynamic shape optimization, and related research on the same type of Π-shaped girder.]]> R² are all larger than 0.99 in the buffering response of the three directions under the high turbulent wind and above 0.98 under the low turbulent wind. Specifically, R² also approaches 0.99 in the torsional angle. Therefore, the proposed TL-Conv LSTM-AM combined model has very high accuracy, with strong generalization in the prediction of the buffeting response of the thin plate. The research conclusions provide new insights into the time-history prediction of the buffeting response of long-span bridges and other engineering structures.]]>