Improved Hilbert-huang Transform Method And Its Application In The Bridge Dynamic Monitoring Data Processing

Through the health monitoring of large bridges to acquire the structure vibration behavior under various conditions is of great significance for the bridges under construction or after it is open to traffic.As an important parameter to reflect the safety status of the bridge structures,the vibration frequency has become the focus in bridge monitoring.As an important method in bridge health monitoring,GPS technology is widely used in the fields of displacement measurement,vibration status monitoring and so on.However,affected by multi-path error and other external environmental factors,GPS dynamic observation data contains lots of noises,which is usually expressed as non-linear,non-stationary.The way of which to excavate the implied extensive structural health information in GPS monitoring signal is very important.In this paper,based on the real-time monitoring system of Sutong Bridge during construction period,dynamic monitoring data are taken as the research project to identify the vibration frequency of bridge structure.For the problem that the precision of Hilbert spectrum analysis is not high with traditional data processing method--Hilbert-Huang Transform(HHT),which caused by the aliasing of the noise signal and the structural vibration signal.An improved HHT method combining wavelet de-noising process is proposed,and the main research work and achievements are as follows:(1)Systematically summarize the progress of GPS technology applied to bridge health monitoring,focus on reviewing some mathematical methods and research status which commonly used to process non-stationary signals.At last,the advantages and disadvantages of those methods such as Short-time Fourier Transform,Wavelet Analysis,and HHT are summarized and analyzed.(2)Outlines the basic principle of Hilbert-Huang Transform.Firstly,the multi-scale decomposition of a set of simulated data to verify the decomposition ability of the HHT method while process non-stationary signals.Then,the components after decomposed are transformed by HT.Through the analysis of those transformation results,discovers the problem that aliasing of the noise signal and the effective information signal in the traditional HHT method.The results show that the Hilbert spectrum analyses are affected by the noise signal,and the existence of the noise affects the accuracy of the results.(3)In order to solve the problem that the precision of Hilbert spectrum analysis is not high in HHT method,an improved HHT method combining wavelet de-noising process is proposed,two de-noising methods based on wavelet threshold de-noising and EMD decomposition are formed.The simulated data and observed dynamic monitoring data are used to comparatively analyze the ability of some methods in signal de-noising.The results show that the Wavelet-EMD method has a good adaptability and it is a suitable method for process non-stationary signals.Using de-noising data for Hilbert spectrum analysis can clearly identify the vibration frequency of the bridge structure,which solves the existent problem in HHT method.(4)Based on the construction of Sutong Bridge,introduce the GPS real-time monitoring system during bridge construction period.With the dynamic monitoring data as research object,the improved HHT method is used for de-noising pretreatment,and the vibration frequency of the bridge structure is identified by using de-noising data.In addition,analyze the change regularity of vibration frequency about the same monitoring point in one day and the correlation of the vibration frequency about different monitoring points in the same construction section.Simultaneously,research the change regularity of the vibration frequency with the construction schedule,which provides favorable support for the safety assessment of bridge structure during construction period.