In this paper, a hybrid FEM-DNN-based vortex-induced vibration (VIV) prediction method for flexible pipes under an oscillatory flow in the time domain is proposed. In this method, a vortex-induced force coefficient model is regressed by a deep neural network (DNN) from experimental data. The model takes into account the effects of flow velocity variation, VIV responses and their coupling features on vortex-induced forces. Then, it is combined with finite element method (FEM) to predict the VIV responses of flexible pipes in time domain. In addition, a phase modulation model is developed to ensure that synchronization between forces and responses can be achieved. The proposed prediction method is used to predict the VIV responses of the flexible pipe used in DNN regression training under oscillatory flows. Comparisons between the predicted results and the experimental results are conducted to verify the feasibility and accuracy of the proposed method. Then, the generalizability of the proposed method is further verified via comparisons between the predicted VIV results and the experimental results of another flexible pipe whose structural parameters are different from the DNN training pipe.

在本文中，提出了一种基于混合FEM-DNN的涡激振动（VIV）预测方法，用于时域振荡流动下的柔性管道。在该方法中，涡激力系数模型通过深度神经网络 (DNN) 从实验数据中回归。该模型考虑了流速变化、VIV 响应及其耦合特征对涡流引起的力的影响。然后，结合有限元法（FEM）对柔性管道的时域VIV响应进行预测。此外，还开发了相位调制模型，以确保可以实现力和响应之间的同步。所提出的预测方法用于预测用于 DNN 回归训练的柔性管在振荡流下的 VIV 响应。将预测结果与实验结果进行对比，验证了所提方法的可行性和准确性。然后，通过将预测的VIV结果与另一个结构参数与DNN训练管不同的柔性管的实验结果进行比较，进一步验证了该方法的普遍性。