基于Hilbert-Huang变换的结构损伤识别研究

发布时间：2018-11-14 13:09

【摘要】：环境激励下结构的振动响应信号一般是非线性、非平稳的,而传统的基于振动的结构损伤识别方法通常建立在系统为线性的前提下。为此,本文利用能自适应地处理非线性和非平稳信号的Hilbert-Huang变换(HHT)和时间序列中的自回归(Auto-regressive,AR)模型进行结构损伤识别研究,主要内容如下:(1)综述了结构损伤识别研究的进展,着重介绍了基于信号处理技术的结构损伤识别方法的发展,指出其应用的范围和存在的问题。(2)简述了HHT及AR模型的基本理论。(3)提出了基于HHT与AR模型的结构损伤识别方法,将加速度响应信号通过经验模态分解(Empirical Mode Decomposition,简称EMD)进行分解,得到单分量的固有模态函数(Intrinsic Mode Function,简称IMF),对其利用AR模型进行拟合,利用结构损伤前后AR模型残差构造损伤特征指标进行结构损伤预警,利用IMF的瞬时相位识别结构的损伤程度。分别在白噪声激励下、非平稳激励下进行数值模拟,并考虑了噪声的影响,最后通过实验验证了该方法的有效性。(4)第四章利用改进的HHT与AR模型结合进行结构的损伤识别。由于信号经EMD分解过程中会产生模态混叠问题。为此,利用Butterworth滤波器将原始的宽带信号过滤为含有单阶频率成分的窄带子信号;为提高损伤预警方法的抗噪性,这里采用窄带子信号的自相关函数进行EMD分解,然后经AR模型对自相关函数进行拟合,利用AR模型的残差构造损伤特征指标进行结构损伤预警,最后利用IMF的瞬时相位识别结构的损伤程度。分别在白噪声激励下、非平稳激励下进行数值模拟,并考虑了噪声的影响,最后通过实验验证改进措施的有效性、合理性。
[Abstract]:The vibration response signals of structures under ambient excitation are generally nonlinear and non-stationary, while the traditional vibro-based structural damage identification methods are usually based on the premise that the system is linear. In this paper, Hilbert-Huang transform (HHT) and autoregressive (Auto-regressive,AR) model in time series are used to study structural damage identification, which can deal with nonlinear and non-stationary signals adaptively. The main contents are as follows: (1) the progress of structural damage identification is reviewed, and the development of structural damage identification based on signal processing is introduced. The application scope and existing problems are pointed out. (2) the basic theory of HHT and AR model is briefly introduced. (3) the structural damage identification method based on HHT and AR model is proposed, and the acceleration response signal is decomposed into (Empirical Mode Decomposition, by empirical mode. EMD) is used to decompose and get the inherent mode function of single component (Intrinsic Mode Function, for short IMF), to fit it by using AR model. The damage characteristic index of AR model before and after structural damage is used to predict structural damage. The instantaneous phase of IMF is used to identify the damage degree of the structure. Numerical simulation is carried out under white noise excitation and non-stationary excitation, and the effect of noise is considered. Finally, the effectiveness of the method is verified by experiments. (4) in chapter 4, the improved HHT and AR model are used to identify the damage of the structure. Because the signal is decomposed by EMD, modal aliasing will occur. For this reason, the original wideband signal is filtered into a narrow band signal with single order frequency by Butterworth filter. In order to improve the noise resistance of the damage warning method, the autocorrelation function of narrow band sub-signal is used to decompose the EMD, and then the autocorrelation function is fitted by the AR model, and the structural damage early warning is carried out by using the residual damage characteristic index of the AR model. Finally, the instantaneous phase of IMF is used to identify the damage degree of the structure. Numerical simulation is carried out under white noise excitation and non-stationary excitation, and the effect of noise is considered. Finally, the effectiveness and rationality of the improved measures are verified by experiments.
【学位授予单位】：青岛理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU317

【相似文献】