大跨斜拉桥北斗监测挠度温度效应分离研究

Research on Deflection Temperature Effect Separation in Beidou Monitoring of Long-span Cable-stayed Bridge

针对大跨斜拉桥北斗监测挠度温度效应分离问题,提出先利用挠度数据的周期特性,在挠度数据首尾分别进行波形延拓,在进行小波分解后所得小波细节中剔除高幅值车载作用挠度得到残余分量,将其与小波系数重构得到预降噪挠度,最后将预降噪挠度进行小波分解来实现挠度数据的高精度降噪;得到降噪挠度后,利用改进的集总平均经验模态分解(MEEMD)良好的可抑制分解过程中产生模态混叠的特性,将降噪挠度进行MEEMD分解,接着将所得日温差和年温差效应第1个半周期通过对称置换得高精度日温差和年温差效应,最后将降噪挠度剔除高精度日温差和年温差效应后所得残余分量再次进行MEEMD分解,所得趋势部分即为长期挠度,从而实现日温差效应、年温差效应、长期挠度的逐步分离。结果表明:波形延拓+预降噪+小波分解的降噪算法比传统单一降噪算法精度更高;温度效应分离算法能实现挠度温度效应各周期成分的精确分离,适合大跨斜拉桥北斗监测挠度温度效应分离。

Aiming at the separation of deflection temperature effect in Beidou monitoring of long-span cable-stayed bridges, the periodic characteristic of deflection data was used to carry out waveform continuation at the beginning and the end of deflection data respectively. The high-amplitude vehicle deflection was removed from the wavelet details after wavelet decomposition and the residual components were gotten. The pre-denoising deflection was obtained by reconstruction of the residual components and wavelet coefficients. The deflection date could be denoised precisely by wavelet decomposition of pre-denoising deflection. After denoising deflection, it could be decomposed by the modified ensemble empirical mode decomposition(MEEMD) because of the fine characteristic that suppressed the mode aliasing in the decomposition process. After MEEMD decomposition, the first half cycle of daily temperature difference and annual temperature difference effect was symmetrically replaced to obtain high precision daily temperature difference and annual temperature difference effect. Finally, the residual components of denoising deflection after eliminating high precision daily temperature difference and annual temperature difference effect were decomposed again by MEEMD. The trend part was long-term deflection, thus the gradual separation of the daily temperature difference effect, the annual temperature difference effect and the long-term deflection were realized. The results show that the denoising algorithms of waveform continuation + pre-denoising + wavelet decomposition is more accurate than the traditional single denoising algorithm. The temperature effect separation algorithm can achieve accurate separation of each periodic component of the deflection temperature effect and it is suitable for deflection temperature effect separation for Beidou monitoring of long-span cable-stayed bridge.