低信噪比下针对平板结构冲击的快速定位算法  

Fast localization algorithm for impact of flat plate structure under low SNR

作　　者：彭涛 崔立林[1,2] 黄秀峰 徐荣武[1,2] 余文晶[1,2] 程果[1,2] PENG Tao;CUI Lilin;HUANG Xiufeng;XU Rongwu;YU Wenjing;CHENG Guo(National Key Laboratory on Ship Vibration and Noise,Naval University of Engineering,Wuhan 430033,China;Institute of Noise and Vibration,Naval University of Engineering,Wuhan 430033,China)

机构地区：[1]海军工程大学振动与噪声研究所,武汉430033 [2]海军工程大学船舶振动噪声重点实验室,武汉430033

出　　处：《振动与冲击》2023年第10期180-187,共8页Journal of Vibration and Shock

基　　金：国家自然科学基金(11904407)。

摘　　要：为实现低信噪比(signal-to-noise ratio,SNR)条件下平板结构受到冲击后的快速准确定位,首先结合集合经验模态分解(ensemble empirical mode decomposition,EEMD)和希尔伯特变换,从冲击信号中提取弯曲波模态分量。利用Hilbert包络线第一峰值作为弯曲波到达时间,将不同传感器到达时间差作为特征构建数据库。利用平板结构中传感器信号能量分布准则判断初始的定位区域。最后计算划分网格的特征残差,实现冲击位置的快速定位。经过平板敲击试验验证,在SNR为5 dB和0的情况下,该算法与基于广义互相关(generalized cross correlation,GCC)和信号阈值的定位算法的定位误差分别为0.08 m,0.17 m,0.18 m和0.13 m,0.26 m,0.28 m,说明该算法具有更好的抗噪性能和更高的定位精度。In order to realize fast and accurate location of flat plate structure after impact under low signal-to-noise ratio(SNR),this paper firstly combines ensemble empirical mode decomposition(EEMD) and Hilbert transform to extract modal component of curved wave from impact signal.The first peak value of Hilbert envelope was used as the arrival time of bending wave,and the arrival time difference of different sensors was used as the feature to construct the database.Secondly,the initial location area was determined by the energy distribution criterion of sensor signal in the plate structure.Finally,the characteristic residuals were calculated to realize the fast location of impact position.It is verified by plate knock experiment that when SNR=5 dB and SNR=0,the positioning errors of the proposed algorithm and the algorithm based on generalized cross correlation(GCC) and signal threshold are 0.08 m,0.17 m,0.18 m and 0.13 m,0.26 m,0.28 m respectively,indicating that the proposed algorithm has better anti-noise performance and higher positioning accuracy.

关 键 词：集合经验模态分解(EEMD) Hilbert包络 时延估计 特征数据库 

分 类 号：TH113.1[机械工程—机械设计及理论]