动载荷识别的物理嵌入式神经网络模型与方法

杨智春; 杨特

doi:10.7527/S1000-6893.2024.31450

航空学报 >

0 1 - 0

DOI: https://doi.org/10.7527/S1000-6893.2024.31450

动载荷识别的物理嵌入式神经网络模型与方法

杨智春 ,
杨特

展开

1. 西北工业大学
2. 西北工业大学航空学院结构动力学与控制研究所

收稿日期: 2024-10-28

修回日期: 2024-12-09

网络出版日期: 2024-12-10

基金资助

航空科学基金;西北工业大学大学博士创新基金;国家自然科学基金

收起

Physical Embedded Neural Network model and method for Dynamic Load identification

YANG Zhi-Chun ,
YANG Te

Expand

Received date: 2024-10-28

Revised date: 2024-12-09

Online published: 2024-12-10

Supported by

Aeronautical Science Foundation of China;Innovation Foundation for Doctor Dissertation of Northwestern Polytechnical University

Fold

摘要

针对传统动载荷识别方法中频响函数矩阵求逆运算导致的不适定性问题，以及深度学习方法缺乏物理可解释性的局限，提出了一种全新的物理嵌入式神经网络（Physical Embedded Neural Network, PENN）动载荷识别模型与方法。通过将结构动力学参数（如模态质量、模态刚度、模态阻尼等）直接嵌入神经网络中，构建出具有物理可解释性的PENN动载荷识别模型。PENN模型能够以正向计算过程直接识别动载荷的功率谱密度，避免了传统方法的频响函数矩阵求逆运算，并能够对其内部的物理参数进行自适应修正，保证了在先验物理参数不准确时仍能实现动载荷的高精度识别。详细阐述了方法机理、PENN模型构建规则、参数设定及训练流程，并对多种工况下的动载荷进行了数值仿真与实验验证，结果表明，本方法在动力学系统先验参数不准确和仅有1组训练样本的情况下，识别动载荷的皮尔逊相关系数均不低于95%，展现出较好的鲁棒性和工程应用潜力。

关键词： 动载荷识别; 物理嵌入式神经网络; 可解释神经网络; 模态分析; 参数校正; 凸优化

本文引用格式

杨智春 , 杨特 . 动载荷识别的物理嵌入式神经网络模型与方法[J]. 航空学报, 0 : 1 -0 . DOI: 10.7527/S1000-6893.2024.31450

Abstract

To address the ill-posedness caused by the inversion of the frequency response function matrix in traditional dynamic load identification methods, as well as the lack of physical interpretability in deep learning methods, a novel Physical Embedded Neural Network (PENN) for dynamic load identification is proposed. By embedding structural dynamic parameters, such as modal mass, modal stiffness, and modal damping, directly into the neural network, the PENN model is constructed to offer physical interpretability. The PENN model can directly identify the power spectral density of dynamic loads through a forward computational process, avoiding the need for inverting the frequency response function matrix as in traditional methods. Additionally, it can adaptively adjust internal physical parameters, ensuring high-precision identification of dynamic loads even when prior physical parameters are inaccurate. The paper provides a detailed explanation of the method’s mechanism, the construction rules of the PENN model, parameter settings, and the training process. Numerical simulations and experimental validations were conducted under various conditions, showing that the Pearson correlation coefficient of dynamic load identification was consistently above 95%, even with inaccurate prior dynamic system parameters and only one training sample, demonstrating strong robustness and potential for engineering applications.

Key words： Dynamic load identification; Physical Embedded Neural Network; Interpretable Neural Network; Modal Analysis; Parameter Correction; Convex Optimization

参考文献

[1]杨智春, 谷迎松, 李斌，等.航空结构动强度分析与设计[M]. 西安：西北工业大学出版社，202108. ISBN: 978-7-5612-7939-7.
[2]王文睿.炮舱结构在冲击载荷作用下的振动特性分析[D]. 西北工业大学, 2007.
[3]赵志彬.飞机炮舱结构有限元静力分析及疲劳寿命估算[D]. 西北工业大学, 2007.
[4]陈忠明, 何连珠.炮舱结构段的动特性与动响应分析[J]. 飞机设计, 1998(03):16-23.[J].飞机设计, 1998, 03(00):16-23
[5]杨智春, 贾有.动载荷的识别方法[J].力学进展, 2015, 45(00):29-54
[6]Bartlett F D, Flannelly W G.Model verification of force determination for measuring vibratory loads[J].Journal of the American Helicopter Society, 1979, 24(2):10-18
[7]张景绘, 李万新.直升机六力素识别[J]. 航空学报, 1986(02):139-147.[J].航空学报, 1986, 02(00):139-147
[8]方明新.超高层建筑风荷载反演分析[D]. 武汉理工大学, 2016.
[9]黄坤.基于高层建筑动力响应的动荷载反演研究[D]. 武汉理工大学, 2019.
[10]Law S S, Chan T H T, Zeng Q H.Moving force identification: a time domain method[J]. Journal of Sound and Vibration, 1997, 201: 1-22.[J].Journal of Sound and Vibration, 1997, 201(00):1-21
[11]张方, 秦远田, 邓吉宏.桥梁结构移动载荷识别的新方法[J]. 南京航空航天大学学报, 2006(01):22-26.[J].南京航空航天大学学报, 2006, 01(00):22-26
[12]ZHANG F., QIN Y. T., DENG J. H. A New Method for Identifying Moving Loads on Bridge Structures [J].Journal of Nanjing University of Aeronautics and Astronautics, 2006, 01(00):22-26
[13]杜玲, 李范春.西部高墩大跨径桥梁桥墩的失稳载荷识别方法[J].应用基础与工程科学学报, 2016, 24(02):304-314
[14]董庆锋, 徐兴平, 畅元江, 上官冬, 吕征.时域载荷识别法在海洋结构中的应用[J]. 钢结构, 2005(05):62-64.[J].钢结构, 2005, 05(00):62-64
[15]Okubo N, Tanabe S, Tatsuno T.Identification of forces generated by a machine under operating condition [C]. Proceedings of the 3rd international modal analysis conference (IMAC). Orlando, FL, 1985, 920-927.
[16]冀灵子.基于电机电流的转子系统载荷识别方法研究[D]. 太原理工大学, 2016.
[17]郑海起, 马吉胜.基于铣削力识别的铣削监测与铣刀故障诊断[J]. 振动、测试与诊断, 2002(02):9-12+67.[J].振动、测试与诊断, 2002, 02(00):9-12
[18]赵继, 王立江, 孟继安.超声振动切削系统中弯曲振动刀杆的谐振模式[J]. 声学学报, 1992(01):22-29.[J].声学学报, 1992, 01(00):22-29
[19]孟光, 周徐斌, 苗军.航天重大工程中的力学问题[J].力学进展, 2016, 46(00):267-322
[20]王磊, 刘亚儒, 胥涵颖.多源不确定性下的飞行器动载荷识别研究进展英文[J].Transactions of Nanjing University of Aeronautics and Astronautics, 2021, 38(02):271-287
[21]张曾, 张江监, 王裕昌.飞机起落架滑行载荷识别[J]. 航空学报, 1994(01):54-61.[J].航空学报, 1994, 01(00):54-61
[22]侯乔乔.全机落震载荷识别方法研究[J].工程与试验, 2019, 59(04):24-26
[23]王洪波, 赵长见, 廖选平, 王亮.基于飞行工作模态分析的飞行器动载荷识别研究[J].动力学与控制学报, 2017, 15(02):178-183
[24]刘恒春, 朱德懋, 孙久厚.振动载荷识别的奇异值分解法[J].振动工程学报, 1990, 3(1):24-33
[25]O' Callahan J, Piergentili F.Force estimation using operational data[C]. Proceedings of the 14rd international modal analysis conference (IMAC). XIV, USA, 1996, 1586-1592.
[26]Karlsson S E S.Identification of external structural loads from measured harmonic responses[J], Journal of Sound and Vibration, 1996, 196: 59-74.[J].Journal of Sound and Vibration, 1996, 196(00):59-74
[27]田燕, 王菁, 郑海起.动载荷识别频响函数矩阵求逆法的改进算法[J].军械工程学院学报, 2002, 14(4):13-17
[28]吕洪彬.基于逆系统的动态载荷识别研究[D]. 大连理工大学，2010.
[29]刘恒春, 朱德懋, 孙久厚.振动载荷识别的奇异值分解法[J].振动工程学报, 1990, 3(1):24-33
[30]O' Callahan J, Piergentili F.Force estimation using operational data[C]. Proceedings of the 14rd international modal analysis conference (IMAC). XIV, USA, 1996, 1586-1592.
[31]Karlsson S E S.Identification of external structural loads from measured harmonic responses[J], Journal of Sound and Vibration, 1996, 196: 59-74.[J].Journal of Sound and Vibration, 1996, 196(00):59-74
[32]田燕, 王菁, 郑海起.动载荷识别频响函数矩阵求逆法的改进算法[J].军械工程学院学报, 2002, 14(4):13-17
[33]吕洪彬.基于逆系统的动态载荷识别研究[D]. 大连理工大学，2010.
[34]Yu L, Chan T H T.Moving force identification based on the frequency-time domain method[J]. Journal of Sound and Vibration, 2003, 261: 329-349.[J].Journal of Sound and Vibration, 2006, 261(00):329-349
[35]Jacquelin E, Bennani A, Hamelin P.Force reconstruction: analysis and regularization of a deconvolution problem[J].Journal of Sound and Vibration, 2003, 265(1):81-107
[36]Liu Y, Shepard Jr W S.Dynamic force identification based on enhanced least squares and total least-squares schemes in the frequency domain[J].Journal of sound and vibration, 2005, 282(1):37-60
[37]毛玉明, 陈建, 刘靖华.考虑模型误差的动载荷反演问题研究[J].振动与冲击, 2012, 31(24):16-19
[38]梅立泉, 崔维庚.面载荷识别的TSVD正则化方法[J]. 应用力学学报, 2010, 27: 140-144.[J].应用力学学报, 2010, 27(00):140-144
[39]毛玉明, 陈建, 刘靖华.考虑模型误差的动载荷反演问题研究[J].振动与冲击, 2012, 31(24):16-19
[40]郑晓霞, 郑锡涛, 缑林虎.多尺度方法在复合材料力学分析中的研究进展[J].力学进展, 2010, 40(01):41-56
[41]Zhou J M, Dong L L, Guan W, Yan J.Impact load identification of nonlinear structures using deep recurrent neural network[J]. Mechanical Systems and Signal Processing, 2019, 133-143.[J].Mechanical Systems and Signal Processing, 2019, 00(00):133-143
[42]Staszewski W J, Worden K, Wardle R, et al.Fail-safe Sensor Distributions for Impact Detection in Composite Materials[J].Smart Materials and Structures, 2000, 9(3):298-303
[43]Ghajari M, Sharif-Khodaei Z, Aliabadi M H.Identification of Impact force for smart composite stiffened panels[J].Smart Materials & Structures, 2013, 22(22):085014-000000
[44]Wang J Y.MIMO SVM based uncorrelated multi-source dynamic random load identification algorithm in frequency domain[J]. Journal of Computational In-formation Systems, 11:22(2015) 8165-8176.[J].Journal of Computational In-formation Systems, 2015, 11(22):8165-8176
[45]Fang X Y, Zhang J, Zhao Y X.Non-linear identification of one-stage spur gearbox based on pseudo-linear neural network[J].Neurocomputing, 2018, 308(9. 25):75-86
[46]Ren S F, Li T G, Chen Q J.A deep learning-based computational algorithm for identifying damage load condition: An Artificial Intelligence Inverse Problem Solution for Failure Analysis[J].Computer Modeling in Engineering & Sciences, 2018, 117(3):287-307
[47]Chen G R, Li T G, Chen Q J.Application of deep learning neural network to identify collision load conditions based on permanent plastic deformation of shell structures[J].Computational Mechanics: Solids, Fluids, Fracture Transport Phenomena and Variational Methods, 2019, 64(2):435-449
[48]夏鹏.利用深度学习的动载荷识别方法研究[D]. 西北工业大学, 2019.
[49]夏鹏, 杨特, 徐江, 王乐, 杨智春.利用时延神经网络的动载荷倒序识别[J].航空学报, 2021, 42(07):389-397
[50]杨特, 杨智春, 梁舒雅, 康在飞, 贾有.平稳随机载荷的信号时频特征提取与深度神经网络识别[J]. 航空学报, DOI: 10. 7527/S1000-6893. 2021. 25952.[J].航空学报, 2021, 00(00):0000-0000
[51]张方, 朱德懋.基于神经网络模型的动载荷识别[J]. 振动工程学报, 1997(02):40-46.[J].振动工程学报, 1997, 02(00):40-46
[52]You Jia, Zhichun Yang, Ning Guo, Le Wang: Random dynamic load identification based on error analysis and weighted total least squares method[J], Journal of Sound and Vibration, 2015, 358: 111-123.[J].Journal of Sound and Vibration, 2015, 358(00):111-123
[53]伍旭强, 张雷, 朱继梅.结构有限元模型的元素修正法[J].振动工程学报, 1992, (02):178-181.DOI:10.16385/j. cnki.issn.1004-4523.1992.02.014.[J].振动工程学报, 1992, 00(00):178-181
[54]朱伯芳.有限单元法原理与应用[M]. 中国水利水电出版社: 201806. 748.
[55]ZHU B.F. Principles and Applications of Finite Element Method [M]. China Water & Power Press, 2018. 748.
[56]Hu H .Vibration Mechanics[M].Springer, Singapor-e:DOI:10.1007/978-981-16-5457-2.

Options

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献