基于LSTM的超临界机翼抖振边界预测方法  

Buffeting boundary prediction method of supercritical wing using LSTM

作　　者：王紫浩 李滚 刘大伟[2] 陈德华 张书俊[1] WANG Zihao;LI Gun;LIU Dawei;CHEN Dehua;ZHANG Shujun(School of Aeronautics and Astronautics,University of Electronic Science and Technology of China,Chengdu 611731,China;High Speed Aerodynamics Institute of China Aerodynamics Research and Development Center,Mianyang 621000,China)

机构地区：[1]电子科技大学航空航天学院,成都611731 [2]中国空气动力研究与发展中心高速空气动力研究所,绵阳621000

出　　处：《空气动力学学报》2024年第6期56-65,I0001,共11页Acta Aerodynamica Sinica

基　　金：旋翼空气动力学重点实验室研究开放课题(2104RAL202102-1)。

摘　　要：超临界机翼的抖振对运输机的安全性和稳定性有着极大的影响,如何高效准确地确定抖振边界一直是备受关注的研究热点。针对CHN-T1型运输机标模,构建了一种基于长短时记忆(long short-term memory, LSTM)神经网络的超临界机翼抖振边界预测框架。根据CHN-T1标模的计算数据,设计了基于LSTM的气动力系数预测模型和抖振起始迎角判定模型,用于准确预测给定马赫数下气动力系数的变化趋势,并且实现了抖振起始迎角的快速判定;通过整合抖振起始迎角数据确定了CHN-T1标模的抖振边界,并用风洞试验数据验证了结果的准确性。研究结果显示,LSTM模型对气动力系数变化趋势有良好的预测能力,其均方根误差维持在2%以内;同时,在抖振起始迎角的判定方面表现出色,抖振边界的误差保持在2%以内。这些结果验证了该方法在抖振边界预测中的可靠性和准确性,为超临界机翼的抖振研究提供了有力支持。The buffeting of supercritical airfoils significantly impacts the safety and stability of transport aircraft.Efficient and accurate determination of buffeting boundaries has been a focal point of research.In this study,a prediction framework for buffeting boundaries of supercritical airfoils was developed using Long Short-Term Memory(LSTM)neural networks,focusing on the CHN-T1 transport aircraft model.Utilizing computational data from the CHN-T1 model,LSTM-based models for predicting aerodynamic coefficients and determining buffeting onset angles were designed.These models forecast changes in aerodynamic coefficients accurately at a given Mach numbers and rapidly determine buffeting onset angles at a given Mach numbers.Integration of buffeting onset angle data ultimately defined the buffeting boundaries of the CHN-T1 model,validated with wind tunnel experimental data.The results demonstrated the LSTM model's excellent predictive capabilities for aerodynamic coefficient trends,maintaining a RMSE within 2%.Furthermore,the model exhibited outstanding performance in buffeting onset angle determination,with errors remaining within 2%.These findings validate the reliability and accuracy of this approach in buffeting boundary prediction,providing robust support for research on supercritical airfoil buffeting.

关 键 词：超临界机翼 抖振边界 气动力系数预测 长短时记忆 

分 类 号：V211.3[航空宇航科学与技术—航空宇航推进理论与工程] O354[理学—流体力学]