| 摘要: |
| 超声导波作为一种非破坏性检测技术，若超声波在有限弹性介质中传播时遇到材料内部缺陷或异物，
会形成反射波和透射波。浸没复杂结构的疲劳裂纹，其反射波与透射波具有典型的时频交叉、非平稳和稀疏性
等特点。传统时频分析方法在处理该类信号时大多存在幅值失真、能量发散、模态混合等问题，导致得到的时
频表示具有较低的时频分辨率，影响裂纹情况的准确识别。为了提高目标模态波包的提取精度、提取时频交叉
信号高分辨率瞬时频率，提出了多重压缩脊线重组法。多重压缩脊线重组法首先从瞬时频率重叠的信号中提取
多分量信号的脊线，通过在短时傅里叶变换中考虑高斯窗提高脊线精度。然后根据脊线在交叉点处的变化率对
脊线进行分组来提取期望的瞬时频率。在获得瞬时频率后，使用本征啁啾分量分解方法实现多分量分离。最后，
通过对时频交叉模拟信号和实验信号的分析，验证了该方法的有效性。 |
| 关键词: 频率压缩 时间压缩 多分量信号 信号分解 |
| 基金项目:国家自然科学基金面上项目（52171283）。 |
|
| Time-frequency Decomposition Method for Sparse Modes |
|
Zhang Wei11, Sun Hailiang22, Zhang Kun1*3, Li Qiang11, Cui Jingzhi11
|
|
1. Beijing Institute of Astronautical Systems Engineering, Beijing 100076;2. China Academy of Launch Vehicle Technology, Beijing 100076;3. College of Mechanical & Energy Engineering, Beijing University of Technology, Beijing 100124
|
| Abstract: |
| Ultrasonic guided wave as a non-destructive detection technology, if ultrasonic waves propagate in a
finite elastic medium when encountering internal defects or foreign objects in the material, it will form reflected and
transmitted waves. The reflected and transmitted waves of fatigue cracks immersed in complex structures are
characterized by typical time-frequency crossover, non-smoothness and sparseness. Traditional time-frequency analysis
methods mostly suffer from amplitude distortion, energy dispersion, and modal mixing when dealing with such signals,
resulting in the obtained time-frequency representations with low time-frequency resolution, which affects the accurate
identification of the crack condition. In order to improve the extraction accuracy of the target modal wave packet and
extract the high-resolution instantaneous frequency of the time-frequency cross-signal, the multiple compression ridge
reorganization method is proposed. The multi-compression ridge reorganization method firstly extracts the ridges of
multi-component signals from the signals with overlapping instantaneous frequencies, and improves the accuracy of the
ridges by considering Gaussian windows in the short-time Fourier transform. The desired instantaneous frequencies are
then extracted by grouping the ridges according to their rate of change at the intersection point. After obtaining the instantaneous frequencies, multi-component separation is achieved using the intrinsic chirp component decomposition
method. Finally, the effectiveness of the method is verified by analyzing the simulated and experimental signals of the
time-frequency crossover |
| Key words: frequency compression time compression multicomponent signal signal decomposition |
