基于全局特征增强图神经网络的降落伞开伞 特性预测

Parachutes possess characteristics like light weight, flexibility and foldability, which are widely employed in fields such as the airdropping of supplies, spacecraft re-entry, and the deceleration and stabilization of ammunition. To obtain parachutes with optimal performance, it is essential to conduct optimized design based on a large number of numerical simulation analyses. However, the existing numerical simulation methods for analyzing parachute opening performances typically take a long time and cannot meet the requirements of rapid iteration in optimized design. Aiming at the problem of the difficulty in quickly predicting parachute opening performances, a graph neural network is utilized to describe the complex parachute structure with multiple materials such as canopies and suspension lines, a rapid prediction model for parachute opening characteristics is constructed, and the refined prediction of the parachute deployment shape is realized. Based on the graph neural network, a global feature enhancement strategy is further introduced to improve the prediction accuracy of the parachute shape. The prediction of the parachute opening dynamic load is realized through the global feature. The analysis results indicate that the graph neural network enhanced by global feature can effectively enhance the prediction accuracy of parachute opening characteristics. The prediction error of the peak value of parachute opening dynamic load and the peak value of projected area for the same type of parachute is less than 10%, and it has a certain generalization ability for parachutes with different structures. This model provides efficient support for parachute optimization design, meeting rapid iteration requirements in engineering applications.