基于核相关滤波器算法的桥墩振动位移及动力特性识别  

Identification of vibration displacement and dynamic characteristics of piers based on the kernelized correlation filters algorithm

作　　者：陈良玉 蔡玮 谢文 何天涛 CHEN Liangyu;CAI Wei;XIE Wen;HE Tiantao(Department of Civil Engineering,Ningbo University,Ningbo 315211,China;Department of Bridge Engineering,Tongji University,Shanghai 200092,China;Ningbo Municipal Facilities Center,Ningbo 315010,China)

机构地区：[1]宁波大学土木工程系,浙江宁波315211 [2]同济大学桥梁工程系,上海200092 [3]宁波市市政设施中心,浙江宁波315010

出　　处：《振动与冲击》2025年第8期267-275,共9页Journal of Vibration and Shock

基　　金：国家自然科学基金(52278189)。

摘　　要：桥梁的振动位移可反映桥梁的力学性能及运营状态,同时通过振动位移可反演桥梁的动力特性,如模态和频率等参数,从而评估桥梁的运营状态和损伤状况,而传统的位移监测技术成本高和测点有限。该研究提出了一种低成本、非接触、多点测的基于核相关滤波器(kernelized correlation filters,KCF)算法的桥梁小幅振动位移视觉测量方法,开展了不同白噪声扫频下双柱式桥墩模型振动台试验,采用激光位移计(laser displacement sensor,LDS)作为参考进行比较验证,利用协方差驱动的随机子空间方法识别了桥梁固有频率及模态振型,验证了采用KCF算法在识别双柱式桥墩乃至桥梁小幅振动位移及相应模态频率的可靠性、可行性和准确性。结果表明:基于KCF算法识别的双柱式桥墩小幅振动位移与LDS记录的波形、变化趋势和峰值几乎一致,其峰值误差在4.0%以内;采用机器视觉识别的振动位移识别的双柱式桥墩固有频率与LDS结果之间的误差在2.5%之内,两者之间识别的模态振型置信水平达0.90以上。Evaluating the mechanical performance and operational conditions of bridge structures relies on accurate measurement of vibration displacement.Such measurement can provide essential parameters,such as mode and frequency,to assess the operational status and condition of damaged bridge structures.However,traditional displacement monitoring techniques have high cost,low accuracy and limited measurement positions.A low-cost,non-contact,and multi-point measurement method was proposed in this paper,based on the kernelized correlation filters(KCF)algorithm to measure the vibration displacement of bridges.Shaking table tests on a dual-column pier model were conducted,with energy dissipation links under different white noise sweeps.The recorded vibration displacement from a laser displacement sensor(LDS)was used as a reference for comparison.The natural frequencies and mode shapes of the dual-column pier and bridges were identified using the covariance-driven stochastic subspace identification method.The reliability,feasibility,and accuracy of machine vision technology in identifying the natural frequencies and mode shapes of bridges were verified.The results show that the small amplitude vibration displacement of the dual-column pier identified by the KCF algorithm is almost consistent with the waveforms,change trends,and peak values recorded by LDS,with a maximum peak error of 4.0%.The error between the natural frequencies identified by the KCF algorithm and the LDS results was within 2.5%.The confidence level of the mode shapes identified between both approaches was above 0.90.

关 键 词：机器视觉 改进的核相关滤波器(KCF)算法 小幅振动位移 动力特性 振动台试验 

分 类 号：U448.34[建筑科学—桥梁与隧道工程]