关键词: 倾转旋翼机, 气弹稳定性, 去噪扩散概率模型, 回转颤振, 多体系统, 数据增强

Abstract: The modal damping ratio and transient response are critical indicators for characterizing the stability of tiltrotor multi-body systems. However, achieving both rapid and accurate predictions of these indicators under multiple operating conditions remains a significant challenge, impeding a comprehensive exploration of the whirl flutter mechanism. To address this gap, this paper proposes a fast prediction framework for tiltrotor aeroelastic stability based on a Denoising Diffusion Probabilistic Model (DDPM). By innovatively integrating dynamic parameters and inflow conditions as conditional inputs, incorporating data augmentation strategies, and embedding an attention mechanism, the framework achieves the unified prediction of the system's modal damping ratio and aeroelastic response. This approach aims to enhance both the efficiency and accuracy of stability predictions for multi-body systems while effectively mitigating the issues of unstable training and low-fidelity results often encountered when traditional generative models handle highly nonlinear problems. First, multi-modal coupled aeroelastic dynamic equations are derived using Hamilton's principle and multi-body dynamics methods, and their validity is verified. Second, a high-fidelity dataset of modal damping ratios and responses is constructed, covering a comprehensive range of dynamic parameter combinations and inflow velocities. Finally, the prediction model is trained with optimized hyperparameters and subjected to performance evaluations. The results demonstrate that the proposed model achieves high accuracy in predicting the stability of tiltrotor multi-body systems and exhibits excellent generalization capability. Compared to traditional numerical calculation models, the prediction efficiency is improved by three to four orders of magnitude, thereby providing robust technical support for research on the whirl flutter mechanism and for the optimization of engineering design.

Key words: tiltrotor aircraft, aeroelastic stability, denoising diffusion probabilistic model (DDPM), whirl flutter, multi-body system, data augmentation