桥梁动力测试信号的自适应分解与重构

针对桥梁结构动力测试信号噪声水平高、难以分离结构有效信号的特点,在总体平均经验模态分解方法和主成分分析的基础上,建立了自适应分解与重构方法。对经验模态分解结果的模态混叠现象进行深入分析,利用白噪声概率密度函数的均匀性对模态混叠模式一进行了改进,基于相关性分析改进了模态混叠模式二,改进后的分解方法在计算效率和分解精度上均有较大提升;随后对所有分解获得的固有模态函数进行多尺度主成分分析,实现降噪和选择并重构测试信号。分别用模拟信号和实际桥梁测试信号对所提方法的有效性进行了验证。结果表明:改进后的信号自适应分解和重构方法能在降噪的同时,有效地提取桥梁结构信息,可用于实际桥梁结构的动力测试分析中。

Adaptive decomposition and reconstruction for bridge structural dynamic testing signals

In order to extract structural information from the bridge structural dynamic signal with high noise level, a novel adaptive decomposition and reconstruction method is proposed by combining the ensemble empirical mode decomposition (EEMD) and principal component analysis (PCA) for the specific characteristics of bridge structural dynamic signals. Based on the in-depth analysis of mode mixing in empirical mode decomposition, the uniformity of probability density function for white noise is adopted to improve the pattern one of mode mixing, and the correlation analysis is used to ameliorate the pattern two of mode mixing, then the calculation efficiency and decomposition accuracy are upgraded greatly in the improved EEMD. The multi-scale principal components analysis is implemented on all of the intrinsic mode functions (IMFs) obtained by the improved EEMD for noise reduction and selection of IMFs. Moreover, the dynamic signal is reconstructed. The effectiveness of the proposed method is verified by both of the simulated signal and testing signal from real bridge structure. The verified results showed that the proposed method can decompose adaptively and denoise effectively the bridge dynamic signal with high noise, and can extract accurately the structural information from the testing signal, furthermore it is applicable in the dynamic testing analysis of real bridge structure.