小波-分形联合的结构损伤无损振动诊断

Nondestructive Vibration Monitoring of Structural Damage Using Wavelet and Fractal

基于小波的强局部分析能力和分形的强非线性处理能力,提出了结构损伤诊断的小波细节关联维分析方法。该方法由两个主要步骤构成:首先采用快速二进小波变换àtrous算法对振动检测波进行多分辨率分析,并利用àtrous算法的可加性重构特性和小波细节频谱保证标准(WDFSAC)确定含损伤信息的关键频段;然后对关键频段内的小波细节进行关联维分析,产生关联积分和关联维计算曲线作为损伤诊断平台,并在关联维计算曲线的灵敏尺度区间内设立关联维阀值判据,实现结构损伤的参数化诊断。通过数值和物理实验对方法进行充分检验,实验结果说明,该方法不依赖模型进行损伤诊断,具有较强的弱损伤诊断能力和噪声免疫能力,构成了一条可行、有效的结构损伤无损振动诊断技术路线。

A new structural damage monitoring method using correlation dimension (D_c) of wavelet details of vibration detection wave to monitor structural damage is proposed on basis of combining the significant ability of wavelet in signal local analysis with that of fractal in nonlinear processing. The method is made up of two main steps. Firstly, the vibration detection wave is decomposed into varying scales of temporal resolution by à trous algorithm of dyadic wavelet transform in order to protrude the signal components on damage, then one or several key frequency bands or scales containing main damage information are determined by the additivity of à trous algorithm and WDFSAC (wavelet detail frequency spectrum assurance criterion). Secondly,D_c analysis of wavelet details of key frequency bands is carried out and the correlation integral curve and D_c computation curve are presented to provide a platform for damage monitoring, then the D_c threshold criterion for implementing parameterized damage monitoring is established at the sensitive scales interval of the D_c computation curve. The proposed approach is fully tested with numerical and physical experiments. The results show its advantages such as being free of computing model, being of high sensitivity to small damage and possessing strong immunity to noise. Therefore it forms a feasible and effective technology routine for nondestructive vibration monitoring on structural damage.