Prediction of Raindrop Vertical Velocity Using a BP Neural Network and Wind–Rain-Induced Dynamic Response Analysis of the FAST Main Supporting Structure

Strong winds are often accompanied by heavy rainfall, and the FAST main supporting structure is sensitive to wind–rain load. Therefore, it is necessary to investigate the wind–rain-induced vibration response of the FAST structure. Given the limited predictive accuracy of existing raindrop vertical velocity models developed through numerical simulations, this study proposes a backpropagation (BP) neural network-based prediction method. In total, 1[Formula: see text]440 sets of raindrop velocity data under varying parameter conditions are extracted from the literature to train the BP neural network, improving both prediction performance and flexibility. Compared with the previously reported nonlinear regression model, the proposed network reduces the root mean square error (RMSE) to below 0.025, indicating improved prediction accuracy. Based on the BP-predicted vertical velocity, the existing rain-load model is updated and then used to calculate the wind–rain load acting on the FAST main supporting structure. The mean values of the wind–rain-induced dynamic responses, including cable-net node displacement, cable stress, and the displacements of both ring beam nodes and latticed columns, are examined under wind directions of [Formula: see text], [Formula: see text], and [Formula: see text]. The results show that when the cosine of the angle between the surface normal vector and the wind direction is positive, wind–rain load increases with rainfall intensity, whereas it generally decreases when the cosine is negative, reflecting the different directional interactions of wind and rain load components. Moreover, for different wind directions, rain-induced displacement and stress account for a substantial portion of the overall structural response. These findings highlight the need to incorporate wind–rain interaction effects in the design of large-span spatial flexible systems.