An indirect method for measuring frequency domain response of satellite internal structure based on convolutional neural network
-
摘要: 针对生产线卫星脉动式快速生产的要求，为解决力学试验步骤烦琐且内部结构响应测量困难的问题，提出基于卷积神经网络（CNN）的卫星内部结构响应深度学习间接测量方法。分别对生产线I型卫星与成熟的II型卫星进行正弦扫频试验，提取结构加速度频域响应信息，并利用基于CNN建立内部结构响应的间接测量模型验证所提出方法的可行性。结果表明该方法对两种型号卫星内部结构响应的间接测量总体精度分别达到了95.8%与96.9%，具有较强的工程应用潜力。Abstract: In view of the requirement of pulsating rapid production for production line satellite, in order to solve the problems of tedious mechanical test steps and difficulty to measure the internal structural response, a deep-learning -based indirect measurement method based on convolutional neural network (CNN) was proposed. A sine sweep test was conducted on a production line satellite I and a mature satellite II, respectively, to extract the frequency domain response of the structural acceleration. Then an indirect model based on CNN for measuring the internal response was established to verify the feasibility of the proposed method. The results show that the indirect measurement accuracy with the method reaches 95.8% and 96.9% for the two types of satellites, respectively, which indicates that the proposed method may have a strong potential for engineering applications.
-
-
![]()
图 2 基于深度学习的内部测点响应间接测量算法框架
Figure 2. A deep-learning based algorithmic framework for indirect measuring of internal measurement point response
![]()
图 3 某2个测点的加速度幅频特性曲线（I型卫星）
Figure 3. Curves of frequency characteristic of acceleration amplitude at two measurement points (satellite I)
![]()
图 4 某2个测点的加速度幅频特性曲线（II型卫星）
Figure 4. Curves of frequency characteristic of acceleration amplitude at two measurement points (satellite II)
![]()
图 5 内外部测点响应相关性图谱
Figure 5. Correlation spectra of responses of internal and external measurement points
![]()
图 7 间接测量与实际测量结果对比（I型卫星）
Figure 7. Comparison between indirect measurement and actual measurement results (satellite I)
![]()
图 8 间接测量与实际测量结果对比（II型卫星）
Figure 8. Comparison between indirect measurement and actual measurement results (satellite II)
![]()
图 9 间接测量与实际测量结果对比及相关性（I型卫星）
Figure 9. Comparison and correlation of indirect and actual measurement results (satellite I)
![]()
图 10 间接测量与实际测量结果对比及相关性（II型卫星）
Figure 10. Comparison and correlation of indirect and actual measurement results (satellite II)
![]()
图 11 I型卫星间接测量数据精度分析结果
Figure 11. Precision analysis results of indirect measurement data for satellite I
![]()
图 12 II型卫星间接测量数据精度分析结果
Figure 12. Precision analysis results of indirect measurement data for satellite II
![]()
图 13 只考虑主振方向与考虑所有方向的间接测量精度对比（I型卫星）
Figure 13. Comparison among indirect measurement accuracy in main vibration direction only and in all directions (satellite I)
-
[1] 金恂叔. 航天器动力学环境试验的发展概况和趋势[J]. 航天器环境工程, 2003, 20(2): 15-21 DOI: 10.3969/j.issn.1673-1379.2003.02.003 JIN X S. The development status and trends of spacecraft dynamic environment testing[J]. Spacecraft Environment Engineering, 2003, 20(2): 15-21 DOI: 10.3969/j.issn.1673-1379.2003.02.003
[2] ZHAN J W, YOU J J, KONG X, et al. An indirect bridge frequency identification method using dynamic responses of high-speed railway vehicles[J]. Engineering Structures, 2021, 243: 112694 DOI: 10.1016/j.engstruct.2021.112694
[3] LI Y J, LI W, LIU H W, et al. Indirect load measurements for large floating horizontal-axis tidal current turbines[J]. Ocean Engineering, 2020, 198: 106945 DOI: 10.1016/j.oceaneng.2020.106945
[4] ZHANG C, QIN J, YANG Q C, et al. Indirect measurement method of inner wall temperature of scramjet with a state observer[J]. Acta Astronautica, 2015, 115: 330-337 DOI: 10.1016/j.actaastro.2015.05.030
[5] 杜建建, 潘贤德, 刘天一. 航空发动机角接触球轴承轴向力间接测量方法[J]. 航空学报, 2022, 43(9): 184-191 DU J J, PAN X D, LIU T Y. Indirect measurement method of axial load of aero-engine angular contact ball bearing[J]. Acta Aeronautica ET Astronautica Sinica, 2022, 43(9): 184-191
[6] NAGI J, DUCATELLE F, CARO G, et al. Max-pooling convolutional neural networks for vision-based hand gesture recognition[C]//2011 IEEE International Conference on Signal and Image Processing Applications (ICSIPA). Kuala Lumpur, Malaysia, 2011
[7] BAGGA P J, MAKHESANA M A, PATEL H D, et al. Indirect method of tool wear measurement and prediction using ANN network in machining process[J]. Materials Today: Proceedings, 2021, 44: 1549-1554 DOI: 10.1016/j.matpr.2020.11.770
[8] 周聪, 朱新坚, 邵孟. 基于改进BP神经网络的甲醇浓度间接测量方法[J]. 电源技术, 2016, 40(1): 89-93 [9] 马兴瑞, 韩增尧, 邹元杰. 航天器力学环境分析与条件设计研究进展[J]. 宇航学报, 2012, 33(1): 1-12 MA X R, HAN Z Y, ZOU Y J. Advances in mechanical environment analysis and conditional design of spacecraft[J]. Journal of Astronautics, 2012, 33(1): 1-12
[10] LECUN Y, BOTTOU L, BENGIO Y. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324 DOI: 10.1109/5.726791
[11] GLOROT X, BORDES A, BENGIO Y. Deep sparse rectifier neural networks[J]. Journal of Machine Learning Research, 2012, 15: 315-323
[12] PISE A A, ALQAHTANI M A, VERMA P, et al. Methods for facial expression recognition with applications in challenging situations[J]. Computational Intelligence and Neuroscience, 2022: 9261438
[13] 李世晓, 杜锦华, 龙云. 基于一维卷积神经网络的机电作动器故障诊断[J]. 电工技术学报, 2022, 37(增刊1): 62-73 LI S X, DU J H, LONG Y. Fault diagnosis of electromechanical actuators based on one-dimensional convolutional neural network[J]. Transactions of China Electrotechnical Society, 2022, 37(Sup 1): 62-73
[14] 王琦, 邓林峰, 赵荣珍. 基于改进一维卷积神经网络的滚动轴承故障识别[J]. 振动与冲击, 2022, 41(3): 216-223 WANG Q, DENG L F, ZHAO R Z. Fault recognition of rolling bearing based on improved 1D convolutional neural network[J]. Journal of Vibration and Shock, 2022, 41(3): 216-223
[15] 李恒, 张氢, 秦仙蓉, 等. 基于短时傅里叶变换和卷积神经网络的轴承故障诊断方法[J]. 振动与冲击, 2018, 37(19): 124-131 LI H, ZHANG Q, QIN X R, et al. Fault diagnosis method for rolling bearings based on short-time Fourier transform and convolutional neural network[J]. Journal of Vibration and Shock, 2018, 37(19): 124-131
下载:
计量
- 文章访问数: 94
- HTML全文浏览量: 23
- PDF下载量: 14
