轻轨锚固螺杆单位脉冲响应信号的稀疏径向基函数网络提取

摘要/Abstract

摘要：

轻轨锚固螺杆是轻轨交通中连接轨道梁和墩台的关键受力部件，脉冲响应函数的提取是其健康监测的关键和核心技术。引入稀疏优化算法取代经典径向基网络的最小二乘法，构建网络的输入输出方案，引入二维DCT基对学习权值矩阵进行稀疏表示并通过正交匹配追踪算法来获得尽可能稀疏的网络训练权值。将单位脉冲激励信号输入该网络即可获得系统的单位脉冲响应函数。实验结果表明，所构建的稀疏径向基网络能准确提取单位脉冲响应函数，对工程应用具有参考价值。

关键词: 径向基函数(RBF)网络, 轻轨锚固螺杆, 正交匹配追踪算法, 脉冲响应函数

Abstract:

Light rail anchoring screw is a key force component for connecting rail girder and piers.The extraction of the impulse response function is the key premise and the core technology to the screw anchor's health monitoring.Sparse optimization solution was introduced to replace the RBF network's classical least square method.The sparse RBF network with input and output calculation scheme was constructed.Two-dimensional DCT-basis was introduced to represent the weight and the orthogonal matching pursuit(OMP) algorithm was used to get sparser network weights training value.A unit excitation signal was taken as the network's input and the output of the network was the unit impulse response function of the system.The experiments show that optimization solution of the RBF network in the paper can accurately extract the impulse response function.The work herein has reference value in engineering applications.

Key words: radial basis function(RBF) network;light rail anchor screw;orthogonal matching pursuit(OMP) algorithm, impulse response function

中图分类号:

TP391.4

杨丹, 周宇, 叶庆卫, 王晓东. 轻轨锚固螺杆单位脉冲响应信号的稀疏径向基函数网络提取[J]. 中国机械工程, 2015, 26(19): 2606-2612.

Yang Dan, Zhou Yu, Ye Qingwei, Wang Xiaodong. Extraction of Unit Pulse Response Function Based on Sparse RBF Network Applyed in Rail Anchor Screw[J]. China Mechanical Engineering, 2015, 26(19): 2606-2612.

参考文献

[1]左鹤声, 彭玉莺. 振动实验模态分析[M]. 北京: 中国铁道出版社, 1995.
[2]林砺宗, 徐晓莉. 单位脉冲响应函数的测量技术[J]. 振动与冲击, 1994, 13(3): 56-60.
Lin Lizong，Xu Xiaoli.The Measurement Technology of Unit Impulse Response Function[J].Journal of Vibration and Shock,1994, 13(3): 56-60.
[3]陈恩伟, 刘正士, 王勇. 提取脉冲响应函数的小波变换法及时域法分析[J]. 振动工程学报, 2005, 18(2): 189-193.
Chen Enwei，Liu Zhengshi，Wang Yong.Wavelet Transform in Time Domain Method Analysis to Extract the Impulse Response Function[J].Journal of Vibration Engineering,2005,18(2):189-193.
[4]陈恩伟, 刘正士, 陆益民, 等. 一种提取系统脉冲响应函数的高效的去噪算法[J]. 振动工程学报, 2006, 18(4): 475-479.
Chen Enwei，Liu Zhengshi，Lu Yimin，et al. A Noising Algorithm of Extraction System Impulse Response Function with Efficient[J].Journal of Vibration Engineering,2006, 18(4): 475-479.
[5]周凯, 王晓东, 叶庆卫, 等. 基于神经网络解卷积的脉冲响应函数提取[J].中南大学学报(自然科学版)2011,42(S1): 67-71.
Zhou Kai， Wang Xiaodong,Ye Qingwei，et al.Extraction of Impulse Response Function Based on Deconvolution of Neural Network[J]. Journal of Central South University (Science and Technology), 2011,42(S1): 67-71.
[6]任远，白广忱. 径向基神经网络在近似建模中的应用研究[J].计算机应用，2009,29(1):115-118.
Ren Yuan，Bai Guangchen.Radial Basis Neural Network in the Application of Approximation Modeling Research[J]. Journal of Computer Applications, 2009,29(1):115-118.
[7]周伟. 基于非负矩阵稀疏分解和径向基神经网络的人脸识别方法[D]. 成都：电子科技大学, 2006.
[8]潘立登, 吴宁川. 径向基函数神经网络正交最小二乘改进算法的实现[J]. 北京化工大学学报, 2002, 29(4): 82-84.
Pan Lideng，Wu Ningchuan.RBF Neural Network Algorithm to Achieve Improved Orthogonal Least Squares[J].Beijing University of Chemical Technology,2002, 29(4): 82-84.
[9]Donoho D L.Compressed Sensing[J].IEEE Transaction Theory,2006,52(4):1289-1306.
[10]杨予昊, 陈卫东, 王东进. 基于 RBF 网络与 OMP 重构算法的空间自旋目标成像[J]. 中国科学技术大学学报, 2012, 41(11): 941-949.
Yang Yuhao，Chen Weidong，Wang Dongjin. Spatial Target Imaging Reconstruction Algorithm with OMP Based on RBF Network[J].University of Science and Technology of China.2012, 41(11): 941-949.
[11]Wu X, Deng W, Dong Y. A Weighted OMP Algorithm for Doppler Superresolution[C]//Antennas & Propagation (ISAP), 2013 Proceedings of the International Symposium on. IEEE, Dayton,2013: 1064-1067.
[12]韩红平. 压缩感知中信号重构算法的研究[D]. 南京：南京邮电大学, 2012.
[13]石光明, 刘丹华, 高大化, 等. 压缩感知理论及其研究进展[J]. 电子学报, 2009, 37(5): 1070-1081.
Shi Guangming， Liu Danhua， Gao Dahua，et al.Compressed Sensing Theory and Its Research Progress[J].Journal of Electronics, 2009, 37(5): 1070-1081.
[14]Ye Qingwei，Sun Yang，Wang Xiaodong，et al. An Improved LLE Algorithm with Sparse Constraint[J]. Journal of Computational and Theoretical Nanoscience, 2013, 10(12):2872-2876.
[15]王丹丹，周宇，叶庆卫. 基于谱聚类的振动多模态信号频谱分割研究与应用[J]. 中国机械工程，2013, 24(13):1719-1723.
Wang Dandan，Zhou Yu，Ye Qingwei. The Research and Application of Spectral Clustering Based on the Vibration Signal Spectrum Division Multimoda[J].China Mechanical Engineering,2013, 24(13):1719-1723.
[16]王琦, 汪同庆, 叶庆卫. 轻轨锚固螺杆振动信号EMD处理研究[J]. 振动与冲击, 2008, 27(6): 178-181.
Wang Qi，Wang Tongqing，Ye Qingwei.Ratment for Vibration Signal of Light-rail's Screw with EMD[J].Journal of Vibration and Shock,2008, 27(6): 178-181.