基于朴素贝叶斯分类器的多螺栓预紧状态识别  

Multi-bolt preloading state identification method based on Naive Bayes classifier

作　　者：张俊树 李丹[1] 任伟新 ZHANG Junshu;LI Dan;REN Weixin(College of Civil Engineering,Hefei University of Technology,Hefei 230009,Anhui,China;Key Laboratory for Resilient Infrastructures of Coastal Cities,Ministry of Education,Shenzhen University,Shenzhen 518060,Guangdong,China)

机构地区：[1]合肥工业大学土木与水利工程学院,安徽合肥230009 [2]深圳大学土木与交通工程学院滨海城市韧性基础设施教育部重点实验室(筹),广东深圳518060

出　　处：《实验力学》2022年第3期369-377,共9页Journal of Experimental Mechanics

基　　金：国家自然科学基金(51778204,51708164);深圳市科创委项目(KQTD20180412181337494,ZDSYS20201020162400001)。

摘　　要：螺栓是机械和土木工程等领域广泛应用的连接件,利用振动声调制(VAM)技术可检测钢结构连接节点的螺栓松动,但现有的相关研究多针对单螺栓连接结构,无法应用于多螺栓结构。针对多螺栓连接结构,本文以梁间的螺栓拼接板为研究对象,采用VAM实验进行特征提取,提出了基于朴素贝叶斯分类器(NBC)的多螺栓预紧状态识别方法。相比于传统的VAM,该方法的高频激励采用线性调频信号,避免了选频的困扰。通过小波变换处理VAM信号,将不同尺度下小波系数的信息熵和均方根(RMS)作为状态指标,进而找到对调制作用敏感的频段,训练NBC;多螺栓连接不同预紧状态的实验验证了方法的可行性与有效性。实验结果表明,NBC能有效识别多螺栓预紧状态,且本文所提出的状态指标具有良好的鲁棒性。Bolts are widely used in mechanical and civil engineering fields.Vibro-acoustic modulation(VAM)can identify bolt-looseness,but the current studies in the bolt connection mainly focus on identification of the single-bolt joints.In light of VAM,a multi-bolt preloading states identification method based on NBC is proposed in this paper.Compared to traditional VAM,the linear swept sine wave is used as a high excitation to avoid the problem of frequency selection.The information entropy and root mean square(RMS)of wavelet coefficients at different scales are extracted as the state index.Then,the frequency band,which is most sensitive to modulation,could be found.To train the residual by NBC,the multi-bolt preloading state can be identified accordingly.The feasibility and reliability of proposed method are verified by the experiments on multi-bolted connections.It is demonstrated that the proposed NBC-based method can identify the multi-bolt preloading states and the method is robust.

关 键 词：振动声调制(VAM) 多螺栓结构 朴素贝叶斯分类器(NBC) 信息熵 均方根(RMS) 

分 类 号：TB523[理学—物理] U24[理学—声学]