Simultaneous de-noising and enhancement method for long-span bridge health monitoring data based on empirical mode decomposition and fractal conservation law

摘要

This paper introduces a new signal-filtering method, which is aimed at processing long-span bridge structural health monitoring (SIIM) data based on empirical mode decomposition (EMD) and fractal conservation law (FCL). The long-span bridge SHM data is selected from vertical vibration data of the Runyang Bridge's main girder. The key idea of this paper is to enhance the frequency amplitude of the target signal and at the same time to weaken the surroundings-caused signal noise by combining methods as EMD and FCL after confirming the long-span bridge's natural frequency. EMD operation consists of decomposing the vibration data by EMD, selection of effective intrinsic mode functions (IMFs) and effective IMF recombination. FCL theory will be applied on the filtering function to process effective IMFs in the frequency domain before recombination. This theory is based on a partial differential equation modified by a nonlocal fractional anti-diffusive term of lower order. The effect of the EMDFCL filter is compared with the EMD method, FCL method and filter strategy that combines EMD and low-pass filtering, respectively. Results show that the EMDFCL filter outperforms the other signal filtering methods in natural frequency recognition of long-span bridge SHM data. in particular, the bridge's natural frequencies distinguished by EMDFCL match well with the calculated frequencies and original fingerprint.