An Unconstrained Structural Dynamic Load Reconstruction Method Based on the Sparse Bayesian Learning Algorithm
edited-by
-
摘要: 为快速准确重构含有未知初始条件的无约束结构外激励，提出了一种基于稀疏Bayes学习算法的荷载重构方法.结合函数拟合的思想建立控制方程，以噪声服从Gauss分布为先验，在Bayes模型中使用快速算法，稀疏重构未知荷载.为合理表达分段拟合中的初始条件，提出了改进的分段拟合手段，以上一分段末状态响应作为可能初始条件，并辅以低阶振型作为初始位移和初始速度的补充.算例以简化运载火箭模型为研究对象，考虑不同等级噪声和不同初始条件表达形式的影响，验证方法的精度和效率.
-
关键词:
- 函数拟合 /
- 稀疏Bayes学习算法 /
- 改进分段拟合 /
- 荷载重构
Abstract: For rapid and exact reconstruction of dynamic loads on unconstrained structures with unknown initial conditions, a dynamic load reconstruction method was proposed based on the sparse Bayesian learning algorithm. With the idea of the function fitting technique, the control equations were built. The noise was assumed to obey the Gaussian distribution, and the fast algorithm was used in the sparse Bayesian learning model. An improved piecewise fitting method was formulated to rationally express the initial conditions in the piecewise fitting, the end state response of the previous segment was used as the possible initial condition, and the low-order vibration modes were applied as the supplement to the initial displacements and initial velocities. The numerical simulations of simplified launch vehicle models prove the accuracy and efficiency of the proposed method, under the effects of different noise levels and different expressions of initial conditions.-
Key words:
- function fitting /
- sparse Bayesian learning algorithm /
- modified piecewise fitting /
- load reconstruction
edited-by1) 我刊编委张亚辉来稿 -
图 2 所提方法荷载重构流程图
Figure 2. The flowchart of the proposed method for load reconstruction
图 5 本文所提方法不同噪声水平下荷载重构结果
Figure 5. Reconstruction results with the proposed method under different noise levels
图 6 15%噪声水平下不同方法的荷载重构结果
Figure 6. Comparison of reconstruction results between different methods with a 15% noise level
图 7 仅使用低阶振型表达初始条件
Figure 7. Only low-order modes used to express the initial conditions
表 1 重构精度与计算时间
Table 1. Reconstruction accuracies and computation times
proposed ref.[24] noise levels 5% 10% 15% 5% 10% 15% f1 1.87% 3.55% 10.08% 3.17% 10.71% 10.98% f2 0.37% 0.67% 1.82% 0.01% 0.20% 0.35% f3 0.99% 1.67% 2.82% 0.65% 6.34% 9.39% computing time 0.68 s 617.02 s 
下载: 导出CSV
表 2 3种重构方式识别结果对比
Table 2. Comparison of identification results of 3 reconstruction methods
force relative error δ/% computing time t/s only low-order modes expressed
initial conditionf1 58.48 622.91 f2 139.66 f3 105.32 modified piecewise fitting f1 2.56 619.23 f2 4.77 f3 6.03 no segmentation f1 11.58 1 203.26 f2 23.57 f3 16.72
下载: 导出CSV
-
[1] 朱华平. 不适定问题的正则化理论及其应用[D]. 硕士学位论文. 武汉: 武汉理工大学, 2007.ZHU Huaping. The regularization theory for ill-posed problems and application[D]. Master Thesis. Wuhan: Wuhan University of Technology, 2007. (in Chinese) [2] 柏恩鹏, 熊向团. 一种新的正则化方法求解热传导方程的侧边值问题[J]. 应用数学和力学, 2021, 42 (5): 541-550. doi: 10.21656/1000-0887.410290BAI Enpeng, XIONG Xiangtuan. A new regularization method for solving sideways heat equations[J]. Applied Mathematics and Mechanics, 2021, 42 (5): 541-550. (in Chinese) doi: 10.21656/1000-0887.410290 [3] 柳冕, 程浩, 石成鑫. 一类非线性时间分数阶扩散方程反问题的变分型正则化[J]. 应用数学和力学, 2022, 43 (3): 341-352. doi: 10.21656/1000-0887.420168LIU Mian, CHENG Hao, SHI Chengxin. Variational regularization of the inverse problem of a class of nonlinear time-fractional diffusion equations[J]. Applied Mathematics and Mechanics, 2022, 43 (3): 341-352. (in Chinese) doi: 10.21656/1000-0887.420168 [4] 彭凡, 马庆镇, 肖健, 等. 自由运行结构动态载荷识别的格林函数法[J]. 动力学与控制学报, 2016, 14 (1): 75-79. https://www.cnki.com.cn/Article/CJFDTOTAL-DLXK201601012.htmPENG Fan, MA Qingzhen, XIAO Jian, et al. Green kernel function approach of load identification for free structures with overall translation[J]. Journal of Dynamics and Control, 2016, 14 (1): 75-79. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-DLXK201601012.htm [5] 周玙, 刘莉, 周思达, 等. 基于应变模态参数的结构瞬态载荷识别方法研究[J]. 振动与冲击, 2019, 38 (6): 199-205. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201906030.htmZHOU Yu, LIU Li, ZHOU Sida, et al. Transient load identification of structural dynamic systems based on strain modal parameters[J]. Journal of Vibration and Shock, 2019, 38 (6): 199-205. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201906030.htm [6] LIU R, HOU Z, WU P, et al. Dynamic load identification for a power battery pack based on a combined regularization algorithm[J]. Journal of Sound and Vibration, 2022, 529: 116928. doi: 10.1016/j.jsv.2022.116928 [7] QIAO B, CHEN X, XUE X, et al. The application of cubic B-spline collocation method in impact force identification[J]. Mechanical Systems and Signal Processing, 2015, 64/65: 413-427. doi: 10.1016/j.ymssp.2015.04.009 [8] 张超东, 黎剑安. 基于增秩Kalman滤波的动态荷载识别和结构响应重构[J]. 应用数学和力学, 2021, 42 (7): 665-674. doi: 10.21656/1000-0887.410252ZHANG Chaodong, LI Jian'an. Dynamic load identification and structural response reconstruction based on the augmented Kalman filter[J]. Applied Mathematics and Mechanics, 2021, 42 (7): 665-674. (in Chinese) doi: 10.21656/1000-0887.410252 [9] 张梓航. 基于字典学习的移动荷载识别[D]. 硕士学位论文. 合肥: 合肥工业大学, 2020.ZHANG Zihang. Moving load identification based on dictionary learning[D]. Master Thesis. Hefei: Hefei University of Technology, 2020. (in Chinese) [10] QIAO B, ZHANG X, WANG C, et al. Sparse regularization for force identification using dictionaries[J]. Journal of Sound and Vibration, 2016, 368: 71-86. [11] ZHANG Z, HE W, REN W. Moving force identification based on learning dictionary with double sparsity[J]. Mechanical Systems and Signal Processing, 2022, 170: 108811. [12] LIU J, LI K. Sparse identification of time-space coupled distributed dynamic load[J]. Mechanical Systems and Signal Processing, 2021, 148: 107177. [13] SAMAGASSI S, KHAMLICHI A, DRIOUACH A, et al. Reconstruction of multiple impact forces by wavelet relevance vector machine approach[J]. Journal of Sound and Vibration, 2015, 359: 56-67. [14] YAN G, SUN H. A non-negative Bayesian learning method for impact force reconstruction[J]. Journal of Sound and Vibration, 2019, 457: 354-367. [15] PRAWIN J, RAMA MOHAN RAO A. An online input force time history reconstruction algorithm using dynamic principal component analysis[J]. Mechanical Systems and Signal Processing, 2018, 99: 516-533. [16] 张开华. 运载火箭时域分段拟合动态载荷识别研究[D]. 硕士学位论文. 大连: 大连理工大学, 2021.ZHANG Kaihua. Dynamic load identification of launch vehicles by time-domain piecewise fitting[D]. Master Thesis. Dalian: Dalian University of Technology, 2021. (in Chinese) [17] PAN C, YU L. Identification of external forces via truncated response sparse decomposition under unknown initial conditions[J]. Advances in Structural Engineering, 2019, 22 (15): 3161-3175. [18] PAN C, YE X, ZHOU J, et al. Matrix regularization-based method for large-scale inverse problem of force identification[J]. Mechanical Systems and Signal Processing, 2020, 140: 106698. [19] PAN C, DENG X, HUANG Z. Parallel computing-oriented method for long-time duration problem of force identification[J]. Engineering With Computers, 2022, 38 (2): 919-937. [20] TIPPING M E. Sparse Bayesian learning and the relevance vector machine[J]. Journal of Machine Learning Research, 2001, 1 (3): 211-244. [21] TIPPING M E, FAUL A C. Fast marginal likelihood maximisation for sparse Bayesian models[C]//Ninth International Workshop on Artificial Intelligence and Statistics. Florida, USA, 2003: 276-283. [22] LAW S S, ZHU X Q. Moving Loads-Dynamic Analysis and Identification Techniques[M]. London: CRC Press, 2011. [23] 朱斯岩, 朱礼文. 运载火箭动态载荷识别研究[J]. 振动工程学报, 2008, 21 (2): 135-139. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDGC200802007.htmZHU Siyan, ZHU Liwen. Dynamic load identification on launch vehicle[J]. Journal of Vibration Engineering, 2008, 21 (2): 135-139. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDGC200802007.htm [24] PAN C, YU L, LIU H, et al. Moving force identification based on redundant concatenated dictionary and weighted l1-norm regularization[J]. Mechanical Systems and Signal Processing, 2018, 98: 32-49. -
图(9) / 表(2)
计量
- 文章访问数: 369
- HTML全文浏览量: 125
- PDF下载量: 52
- 被引次数: 0
渝公网安备50010802005915号