中国航天空气动力技术研究院 空气动力研究所, 北京 100086
刘国东(1992-), 男, 硕士, 工程师。 E-mail: rendouzao@163.com
收稿:2026-04-01,
修回:2026-07-10,
网络首发:2026-07-29,
移动端阅览
张之豪,刘国东,胡强,等. 基于LSTM的暂冲式风洞马赫数建模与控制方法[J]. 航空工程进展.
ZHANG Zhihao,LIU Guodong,HU Qiang, et al. A LSTM-based modeling and control method for Mach number in an intemitent wind tunnel[J]. Advances in Aeronautical Science and Engineering.(in Chinese)
进行亚跨声速风洞试验时,模型迎角快速变化导致马赫数剧烈振荡,传统PID控制策略难以使马赫数快速稳定。基于长短时记忆神经网络(LSTM),提出一种适用于某暂冲式风洞的马赫数建模与智能控制一体化方法。建模方法使用同一试验模型的历史试验数据对LSTM网络进行训练,以过去20个时间步的扰动(包括气源压力、模型迎角、阀门开度、马赫数)作为输入,预测当前时刻试验段马赫数的输出,建模精度达
R
2
>
0.98。基于LSTM模型,提出小扰动最优化智能控制方法,该方法在每个时间步对阀门开度进行局部扰动寻优,使下一时刻的马赫数输出逼近目标值。结果表明:在理想情况下,目标马赫数1.2、0.95、0.8时马赫数均方根误差(RMS)分别降低76%、82%、51%;即使在引入LSTM模型预测误差后,RMS仍分别降低61%、56%、42%,系统抗扰动能力得到显著提升。
An integrated method for Mach number modeling and intelligent control is proposed based on the Long Short-Term Memory (LSTM) neural network in an intermittent wind tunnel, considering the challenge that rapid changes in the angle of attack of a test model cause severe Mach number oscillations and traditional PID control strategies fail to stabilize the Mach number quickly. The LSTM network is trained using historical test data obtained from the same test model to predict the current Mach number in the test section, with inputs of disturbance variables from the past 20 time steps, including air source pressure, angle of attack of the test model, valve ope-ning degree, and Mach number. The prediction pre
cision of the model achieves
R
2
>
0.98. Based on the established LSTM model, a small-perturbation-optimization based intelligent control method is developed. At each time step, the valve opening degree is locally perturbed and optimized to drive the predicted Mach number in the next time step moving towards the target value. Simulation results demonstrate that the root mean square error (RMSE) of Mach number are reduced by 76%, 82% and 51% at target Mach numbers of 1.2, 0.95 and 0.8, respectively, under ideal conditions, while reduced by 61%, 56% and 42% considering LSTM prediction errors, indicating significantly enhanced anti-disturbance capability.
Anderson J D . Fundamentals of aerodynamics [M]. 5th ed . New York : McGraw-Hill Higher Education , 2011 .
王晓军 . 基于大数据的风洞马赫数集成建模方法的研究 [D]. 沈阳 : 东北大学 , 2016 .
Wang Xiaojun . The research of ensemble methods based on big data for the Mach number prediction in wind tunnel [D]. Shenyang : Northeastern University , 2016 . (in Chinese)
袁平 , 易凡 , 肖宇航 , 等 . 面向攻角变化的风洞流场模型预测控制器 [J]. 控制与决策 , 2018 , 33 ( 6 ): 1026 - 1032 .
Yuan Ping , Yi Fan , Xiao Yuhang , et al . Orienting of attack angle based model prediction controller of wind tunnel flow [J]. Control and Decision , 2018 , 33 ( 6 ): 1026 - 1032 . (in Chinese)
易凡 , 李欣蕊 , 杜宁 , 等 . 基于迭代学习的风洞马赫数控制方法 [J]. 控制工程 , 2020 , 27 ( 1 ): 109 - 113 .
Yi Fan , Li Xinrui , Du Ning , et al . Iterative learning based control for wind tunnel Mach number [J]. Control Engineering of China , 2020 , 27 ( 1 ): 109 - 113 . (in Chinese)
朱天成 . 连续变姿态角工况下风洞马赫数的控制方法研究 [D]. 沈阳 : 东北大学 , 2021 .
Zhu Tiancheng . Research on control method of Mach number in wind tunnel with continuous variable attitude angle [D]. Shenyang : Northeastern University , 2021 . (in Chinese)
刘为杰 , 凌忠伟 , 田嘉懿 , 等 . 2 m高速自由射流风洞流场前馈—反馈复合控制方法研究与应用 [J]. 西北工业大学学报 , 2025 , 43 ( 3 ): 574 - 581 .
Liu Weijie , Ling Zhongwei , Tian Jiayi , et al . Feedforward-feedback compound control method and application for flow field in 2 meter high-speed free jet wind tunnel [J]. Journal of Northwestern Polytechnical University , 2025 , 43 ( 3 ): 574 - 581 . (in Chinese)
刘为杰 , 凌忠伟 , 邓晓曼 , 等 . 风洞流场抗时变干扰控制研究 [J]. 实验流体力学 , 2025 , 39 ( 5 ): 53 - 59 .
Liu Weijie , Ling Zhongwei , Deng Xiaoman , et al . Research on anti time-varying disturbance control of wind tunnel flow field [J]. Journal of Experiments in Fluid Mechanics , 2025 , 39 ( 5 ): 53 - 59 . (in Chinese)
张廷丰 . 大型跨声速风洞建模与控制方法研究 [D]. 沈阳 : 东北大学 , 2019 .
Zhang Tingfeng . Research on modeling and control methods for large transonic wind tunnel [D]. Shenyang : Northeastern University , 2019 . (in Chinese)
Yu W S , Du N , Rao Z Z , et al . Perturbation analysis and control of Mach number 2.4-meter transonic wind tunnel [J]. Journal of Aircraft , 2020 , 57 ( 6 ): 1148 - 1155 .
金志伟 , 杜宁 , 邢盼 , 等 . 基于前馈—模糊PID策略的风洞控制器应用 [J]. 兵工自动化 , 2024 , 43 ( 7 ): 48 - 51 .
Jin Zhiwei , Du Ning , Xing Pan , et al . Application of wind tunnel controller based on feedforward-fuzzy PID strategy [J]. Ordnance Industry Automation , 2024 , 43 ( 7 ): 48 - 51 . (in Chinese)
马靖雯 , 张廷丰 , 陆明超 . 基于数据驱动的跨声速风洞控制方法研究 [J]. 辽宁工业大学学报(自然科学版) , 2024 , 44 ( 5 ): 298 - 302, 309 .
Ma Jingwen , Zhang Tingfeng , Lu Mingchao . Research on data-driven control methods for transonic wind tunnels [J]. Journal of Liaoning University of Technology (Natural Science Edition) , 2024 , 44 ( 5 ): 298 - 302, 309 . (in Chinese)
陈旦 , 王众 , 鲁相 , 等 . 某连续式超声速风洞控制系统设计研究 [J]. 西北工业大学学报 , 2022 , 40 ( 1 ): 167 - 174 .
Chen Dan , Wang Zhong , Lu Xiang , et al . The control system design and study for continuous supersonic wind tunnel [J]. Journal of Northwestern Polytechnical University , 2022 , 40 ( 1 ): 167 - 174 . (in Chinese)
尼文斌 , 董金刚 , 刘书伟 , 等 . 自适应遗传PID算法在风洞风速控制中的应用 [J]. 实验流体力学 , 2015 , 29 ( 5 ): 84 - 89 .
Ni Wenbin , Dong Jingang , Liu Shuwei , et al . Application of PID based on adaptive genetic algorithms in wind velocity control system of wind tunnels [J]. Journal of Experiments in Fluid Mechanics , 2015 , 29 ( 5 ): 84 - 89 . (in Chinese)
Yu W S , Su B C , Rao Z Z , et al . Genetic algorithm-based Mach number control of multi-mode wind tunnel flow fields [J]. Processes , 2022 , 10 ( 10 ): 2038 .
高赫 . 基于机器学习的连续式风洞马赫数控制 [D]. 南京 : 南京航空航天大学 , 2020 .
Gao He . Mach number control of continuous wind tunnel based on machine learning [D]. Nanjing : Nanjing University of Aeronautics and Astronautics , 2020 . (in Chinese)
王晓军 , 袁平 , 毛志忠 , 等 . 基于随机森林的风洞马赫数预测模型 [J]. 航空学报 , 2016 , 37 ( 5 ): 1494 - 1505 .
Wang Xiaojun , Yuan Ping , Mao Zhizhong , et al . Wind tunnel Mach number prediction model based on random forest [J]. Acta Aeronautica et Astronautica Sinica , 2016 , 37 ( 5 ): 1494 - 1505 . (in Chinese)
王凯文 . 神经网络PID控制在FD12风洞亚跨流场校测中的应用 [D]. 哈尔滨 : 哈尔滨工程大学 , 2011 .
Wang Kaiwen . The application of neural network PID control in the test of FD12Supsonic and transonic wind tunnel [D]. Harbin : Harbin Engineering University , 2011 . (in Chinese)
Zhao L P , Wu K Y . Mach number prediction for a wind tunnel based on the CNN-LSTM-attention method [J]. Instrumentation , 2023 ( 4 ): 6 - 11 .
Yu W S , Su B C , Liu G , et al . Mach number prediction of wind tunnel flow field based on RBFNN and LSTM [C]∥ 2024 36th Chinese Control and Decision Conference (CCDC) . Xi’an, China : IEEE , 2024 : 2462 - 2466 .
Zhao L P , Wang C . Physical information-based Mach number prediction and model migration in continuous wind tunnels [J]. Aerospace , 2025 , 12 ( 8 ): 701 .
Liu L , Mao Z Z . Hybrid deep LSTM-GAT network with mechanism information for prediction of Mach number [J]. IEEE Transactions on Instrumentation and Measurement , 2025 , 74 : 2514813 .
Sutcliffe P , Rennie M R . Neural network model predictive control of wind tunnel test conditions [C]∥ 54th AIAA Aerospace Sciences Meeting . San Diego, California, USA : AIAA , 2016 : 1150 - 1162 .
Tian J Y , Ling Z W , Liu W J , et al . Flow field control for 2-meter high-speed free-jet wind tunnel [J]. Chinese Journal of Aeronautics , 2023 , 36 ( 10 ): 77 - 89 .
Yu Y , Si X S , Hu C H , et al . A review of recurrent neural networks: LSTM cells and network architectures [J]. Neural Computation , 2019 , 31 ( 7 ): 1235 - 1270 .
Hochreiter S , Schmidhuber J . Long short-term memory [J]. Neural Computation , 1997 , 9 ( 8 ): 1735 - 1780 .
0
浏览量
0
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010602201714号