APPLICATION OF RECURRENT NEURAL NETWORK IN RESEARCH OF INTELLIGENT WIND TUNNEL BALANCE
-
摘要: 激波风洞地面试验对高超声速飞行器高焓气动特性研究至关重要, 而高精度气动力测量是其中的关键技术. 在脉冲型激波风洞中进行测力试验时, 风洞起动时流场瞬间建立, 对测力系统会产生较大的冲击. 测力系统在瞬时冲击作用下受到激励, 系统的惯性振动信号在短时间内无法快速衰减, 天平的输出信号中会包含惯性振动干扰量, 导致脉冲型风洞测力试验精准度的进一步提高遇到瓶颈. 为了解决短试验时间内激波风洞快速准确测力问题, 发展高精度的动态校准技术是提升受惯性干扰天平性能的关键方法. 因此, 本文采用循环神经网络对天平动态校准数据进行训练和智能处理, 旨在消除输出动态信号中的振动干扰信号. 本文对该方法进行了误差分析, 验证了该方法的可靠性, 并将该方法应用于激波风洞测力试验中, 切实有效降低了惯性振动对天平输出信号的干扰影响. 根据智能模型的样本验证分析, 各分量载荷相对误差比较小, 其中高频轴向力分量处理结果的相对误差约1%. 在风洞试验数据验证中, 也得到了比较理想的结果, 同时与卷积神经网络模型处理的结果进行了对比分析.Abstract: The shock tunnel ground test is vitally important to the research of the high-enthalpy aerodynamic characteristics of hypersonic vehicles, and the high-accuracy aerodynamic measurement is the key technology. When a force measurement test is conducted in an impulse shock tunnel, the flow field is established instantly after the starting process of shock tunnel, at this time, the great impact loads are acting on the force measurement system. The force measurement system is excited under the action of instantaneous impact, and the inertial vibration signal of the system cannot be rapidly attenuated during the short test time. The output signal of the balance will contain the interference due to the inertial vibration, which leads to a bottleneck in the further improvement of the accuracy of the transient force test. In order to improve the force measurement accuracy in the short-duration shock tunnel, the development of high-accuracy dynamic calibration technology is the key method to improve the performance of balance affected by inertial interference. Therefore, in this paper, recurrent neural network is used to train and intelligently process the balance dynamic calibration data, aiming to eliminate the vibration interference signals in the output dynamic signals. The error analysis of the current method is carried out, and the reliability of the current method is verified. The method is applied to the data processing of force test obtained in shock tunnel, and the effect of inertial vibration on the output signal of the balance is effectively reduced. According to the sample verification analysis of the intelligent model, the relative error of each component load is relatively small, where the case of high-frequency axial force component is about 1%. In the verification of wind tunnel force test data, the good results are also obtained, which are compared with those processed by the convolutional neural network model.
-
图 4 轴向力通道输入信号与经过网络模型处理后信号的对比
Figure 4. Comparison of the input signal and validation data by training model (axial force)
图 5 法向力通道输入信号与经过网络模型处理后信号的对比
Figure 5. Comparison of the input signal and validation data by training model (normal force)
图 6 俯仰力矩通道输入信号与经过网络模型处理后信号的对比
Figure 6. Comparison of the input signal and validation data by training model (pitching moment)
图 9 轴向力通道风洞试验信号与模型处理的信号
Figure 9. Wind tunnel test signal and model processed signal (axial force)
图 10 法向力通道风洞试验信号与模型处理的信号
Figure 10. Wind tunnel test signal and model processed signal (normal force)
图 11 俯仰力矩通道风洞试验信号与模型处理的信号
Figure 11. Wind tunnel test signal and model processed signal (pitching moment)
表 1 LSTM和Bi-LSTM模型的相对误差
Table 1. Relative error of LSTM and Bi-LSTM model
Components $\bar F$ $\bar F_{\rm{L}}^*$ $\bar F_{\rm{B}}^*$ ${\delta _{\rm{L}}}$ ${\delta _{\rm{B}}}$ normal force/N 5.943 6.081 5.947 2.32% 0.07% pitching moment/(N·m) 2.409 2.450 2.403 1.70% −0.25% axial force/N 4.699 4.683 4.6997 −0.34% 0.01% 
下载: 导出CSV
表 2 RNN与CNN模型处理数据的相对误差
Table 2. The relative errors of RNN and CNN models in processing data
Coefficients CNN LSTM Bi-LSTM ${\delta _{\rm{L}}}$ ${\delta _{\rm{B}}}$ normal force 0.155 0.160 0.157 3.23% 1.27% pitching moment 0.107 0.103 0.109 −3.74% 1.87% axial force 0.108 0.111 0.106 2.78% −1.85%
下载: 导出CSV
-
[1] 宗群, 曾凡琳, 张希彬等. 高超声速飞行器建模与模型验证. 北京: 科学出版社, 2016.(Zong Qun, Zeng Fanlin, Zhang Xibin, et al. Modeling and Model Verification of Hypersonic Aircraft. Beijing: Science Press, 2016 (in Chinese)) [2] Bernstein L. Force measurement in short-duration hypersonic facilities. AGARDograph No. 214, 1975. [3] 黄志澄. 高超声速气动试验的新进展. 气动实验与测量控制, 1993, 7(1): 1-13 (Huang Zhicheng. The new progress of hypersonic aerodynamic and aerothermodynamic testing. Aerodynamic Experiment and Measurement &Control, 1993, 7(1): 1-13 (in Chinese) [4] Störkmann V, Olivier H, Gronig H. Force measurements in hypersonic impulse facilities. AIAA Journal, 2015, 36(3): 342-348 [5] 艾迪, 许晓斌, 王雄. 风洞天平动态特性多阶惯性补偿技术研究. 实验流体力学, 2018, 32(4): 87-92 (Ai Di, Xu Xiaobin, Wang Xiong. Investigation of wind tunnel balance dynamic characteristics’ multi-order inertial compensation. Journal of Experiments in Fluid Mechanics, 2018, 32(4): 87-92 (in Chinese) [6] Duryea GR, Martin JF. An improved piezoelectric balance for aerodynamic force measurements//IEEE/G-AES 2nd International Congress on Instrumentation in Aerospace Simulation Facilities, Stanford University, Aug. 29-31, Stanford, California, 1966. [7] 湛华海, 张旭, 吕治国等. 一种单矢量风洞天平校准系统设计. 实验流体力学, 2014, 28(1): 70-74 (Zhan Huahai, Zhang Xu, Lü Zhiguo, et al. Design of a single vector wind tunnel balance calibration system. Journal of Experiments in Fluid Mechanics, 2014, 28(1): 70-74 (in Chinese) [8] Sheeran WJ, Duryea GR. The application of the accelerometer force balance in short-duration testing//AIAA 4th Aerodynamic Testing Conference, Apr. 28-30, Cincinnati, 1969 [9] Joarder R, Jagadeesh G. A new free floating accelerometer balance system for force measurements in shock tunnels. Shock Waves, 2003, 13(5): 409-412 doi: 10.1007/s00193-003-0225-y [10] Sahoo N, Mahapatra DR, Jagadeesh G, et al. An accelerometer balance system for measurement of aerodynamic force coefficients over blunt bodies in a hypersonic shock tunnel. Measurement Science and Technology, 2003, 14(3): 260 doi: 10.1088/0957-0233/14/3/303 [11] Saravanan S, Jagadeesh G, Reddy KPJ. Aerodynamic force measurement using 3-component accelerometer force balance system in a hypersonic shock tunnel. Shock Waves, 2009, 18(6): 425-435 doi: 10.1007/s00193-008-0172-8 [12] Simmons JM, Sanderson SR. Drag balance for hypervelocity impulse facilities. AIAA Journal, 1991, 29(12): 2185-2191 doi: 10.2514/3.10858 [13] Mee DJ, Daniel W, Simmons JM. Three-component force balance for flows of millisecond duration. AIAA Journal, 2015, 34(3): 590-595 [14] Robinson MJ, Mee DJ, Tsai CY, et al. Three-component force measurements on a large scramjet in a shock tunnel. Journal of Spacecraft and Rockets, 2004, 41(3): 416 [15] Robinson MJ, Schramm JM, Hannemann K. Design and implementation of an internal stress wave force balance in a shock tunnel. CEAS Space Journal, 2010, 1(1): 45-57 [16] Marineau EC, MacLean M, Mundy EP, et al. Force measurements in hypervelocity flows with an acceleration-compensated strain-gauge balance. Journal of Spacecraft and Rockets, 2012, 49(3): 474-482 doi: 10.2514/1.A32041 [17] Wang YP, Liu YF, Luo CT, et al. Force measurement using strain-gauge balance in a shock tunnel with long test duration. Review of Scientific Instruments, 2016, 87(5): 1068 [18] Wang YP, Liu YF, Jiang ZL. Design of a pulse-type strain gauge balance for a long-test-duration hypersonic shock tunnel. Shock Waves, 2016, 26(6): 835-844 doi: 10.1007/s00193-015-0616-x [19] 汪运鹏, 刘云峰, 苑朝凯等. 长试验时间激波风洞测力技术研究. 力学学报, 2016, 48(3): 545-556 (Wang Yunpeng, Liu Yunfeng, Yuan Chaokai, et al. Study on force measurement in long-test duration shock tunnel. Chinese Journal of Theoretical and Applied Mechanics, 2016, 48(3): 545-556 (in Chinese) doi: 10.6052/0459-1879-15-295 [20] 汪运鹏, 李小刚, 姜宗林. 脉冲型天平高精度全自动校准系统. 中国科学: 物理学 力学 天文学, 2020, 50(6): 76-86 (Wang Yunpeng, Li Xiaogang, Jiang Zonglin. High-accuracy fully automatic calibration system for impulse balance. Science China Physics,Mechanics & Astronomy, 2020, 50(6): 76-86 (in Chinese) [21] 汪运鹏, 杨瑞鑫, 聂少军等. 基于深度学习技术的激波风洞智能测力系统研究. 力学学报, 2020, 52(5): 1304-1313 (Wang Yunpeng, Yang Ruixin, Nie Shaojun, et al. Deep-learning-based intelligent force measurement system using in a shock tunnel. Chinese Journal of Theoretical and Applied Mechanics, 2020, 52(5): 1304-1313 (in Chinese) [22] 杨双龙. 风洞应变天平动态特性与动态校正方法研究. [博士论文]. 合肥: 合肥工业大学, 2014.(Yang Shuanglong. Studies on dynamic characteristics and dynamic correction methods for wind tunnel strain gauge balance. [PhD Thesis]. Hefei: Hefei University of Technology, 2014 (in Chinese)) [23] 徐科军, 朱志能, 李成等. 六维腕力传感器阶跃响应的实验建模. 机器人, 2000, 22(4): 251-255 (Xu Kejun, Zhu Zhineng, Li Cheng, et al. Experimental modeling of six-axis wrist force sensor based on step responses. Robot, 2000, 22(4): 251-255 (in Chinese) doi: 10.3321/j.issn:1002-0446.2000.04.003 [24] 郑红梅, 刘正士. 机器人六维腕力传感器动态性能标定系统的研究. 电子测量与仪器学报, 2006, 20(3): 88-92 (Zheng Hongmei, Liu Zhengshi. Research on the dynamic performance calibration system for robot’s 6-axis wrist force sensor. Journal of Electronic Measurement and Instrument, 2006, 20(3): 88-92 (in Chinese) [25] 刘广孚, 张为公. 车轮力传感器的侧向力动态标定方法. 仪表技术与传感器, 2010(3): 100-103 (Liu Guangfu, Zhang Weigong. Research on dynamic calibration method of lateral force of wheel force transducer. Instrument Technique and Sensor, 2010(3): 100-103 (in Chinese) doi: 10.3969/j.issn.1002-1841.2010.03.035 [26] 郑泽宇, 梁博文, 顾思宇. TensorFlow: 实战Google深度学习框架(第2版). 北京: 电子工业出版社, 2018(Zheng Zeyu, Liangbowen, Gu Siyu. TensorFlow: Practical Google Deep Learning Framework (Version 2). Beijing: Publishing House of Electronics Industry, 2018 (in Chinese)) [27] Bengio Y, Simard P, Frasconi P. Learning long-term dependencies with gradient descent is difficult. IEEE Transactions on Neural Networks and Learning Systems. 1994, 5(2): 157-166 [28] 刘云峰, 汪运鹏, 苑朝凯等. JF-12长实验时间激波风洞10°尖锥气动力实验研究. 气体物理, 2017, 2(2): 1-7 (Liu Yunfeng, Wang Yunpeng, Yuan Chaokai, et al. Aerodynamic force measurements of 10° half-angle cone in JF12 long-test-time shock tunnel. Physical of Gases, 2017, 2(2): 1-7 (in Chinese) [29] Yu H, Esser B, Lenartz M, et al. Gasrous detonation driver for a shock tunnel. Shockwaves, 1992, 2: 245-254 [30] 姜宗林, 李进平, 赵伟等. 长试验时间爆轰驱动激波风洞技术研究. 力学学报, 2012, 44(5): 824-831 (Jiang Zonglin, Li Jinping, Zhao Wei, et al. Investigation into techniques for extending the test-duration of detonation-driven shock tunnel. Chinese Journal of Theoretical and Applied Mechanics, 2012, 44(5): 824-831 (in Chinese) doi: 10.6052/0459-1879-12-160 [31] 李进平, 冯珩, 姜宗林. 激波/边界层相互作用诱导的激波风洞试验气体污染问题. 力学学报, 2008, 40(3): 289-296 (Li Jinping, Feng Heng, Jiang Zonglin. Gas contamination induced by the interaction of shock/boundary layer in shock tunnel. Chinese Journal of Theoretical and Applied Mechanics, 2008, 40(3): 289-296 (in Chinese) doi: 10.3321/j.issn:0459-1879.2008.03.001 [32] 李进平, 冯珩, 姜宗林等. 爆轰驱动激波管缝合激波马赫数计算. 空气动力学学报, 2008, 26(3): 291-296 (Li Jinping, Feng Heng, Jiang Zonglin, et al. Numerical computation on the tailored shock Mach numbers for a hydrogen-oxygen detonation shock tube. Acta Aerodynamica Sinica, 2008, 26(3): 291-296 (in Chinese) doi: 10.3969/j.issn.0258-1825.2008.03.004 -
下载:
点击查看大图
图(11) / 表(2)
计量
- 文章访问数: 748
- HTML全文浏览量: 259
- PDF下载量: 48
- 被引次数: 0
