| As a key transportation hub and control node,bridges are one of the country’s important infrastructures.Their operation monitoring and maintenance require the support and guarantee of high-precision deformation monitoring.With the improvement of the Global Navigation Satellite System(GNSS),it has been widely used in the field of high-precision displacement monitoring.The processing of deformation data requires reliable analysis methods and high accuracy,and GNSS deformation monitoring is susceptible to multipath effects and receiver background noise,resulting in low accuracy of deformation monitoring data.Local Mean Decomposition(LMD)is a new time-frequency analysis method that can adaptively perform multi-scale decomposition and processing for non-stationary and multi-component signals.It draws on the advantages of the Empirical Mode Decomposition(EMD)method,some of the existing problems have been improved theoretically.Based on the GNSS deformation monitoring data of Poyang Lake Second Bridge,this paper studies GNSS monitoring data preprocessing,data noise reduction based on EMD and improved LMD methods,and modal parameter extraction.On this basis,GNSS＿DNR is developed: GNSS bridge deformation monitoring data noise reduction software,the main tasks are as follows:(1)A systematic overview of the research background and application of GNSS technology in the field of bridge deformation monitoring and the development status of deformation data noise reduction methods,and on this basis,the research content of this article is proposed.Summarized the basic principles and algorithm implementation process of the EMD and LMD methods,and compared the noise reduction capabilities of the LMD and EMD methods through simulated data.The results show that the LMD method has a better noise reduction effect than the traditional EMD method.(2)In view of the partial signal-to-noise aliasing in the traditional LMD noise reduction method and the direct classification of the product function(PF)component into the highfrequency noise term to eliminate,causing some real signals to be "submerged",this paper proposes an improved LMD method for multiple decomposition and noise reduction.This method first removes the first PF component obtained by LMD decomposition,and then reconstructs the second PF component to the boundary PF component as a high-frequency signal,and performs the next LMD decomposition of the high-frequency reconstructed signal to obtain useful information.Repeat this process many times,and finally accumulate all the obtained low-frequency signals to achieve the purpose of noise reduction.Several sets of simulation data are designed to verify the effectiveness of the method.The visual display of the image after noise reduction and the evaluation index of noise reduction quantitatively show that the noise reduction performance of the improved LMD noise reduction algorithm is better than that of the traditional LMD noise reduction and EMD noise reduction methods.(3)Based on the research of(1)and(2),designed and developed GNSS＿DNR: GNSS bridge deformation monitoring data noise reduction software,which can realize GNSS data preprocessing,data noise reduction based on EMD,LMD and its improved methods,time series diagram drawing,statistical analysis and spectrum analysis and other functions.Combining three sets of simulation data with different types of signals and adding different noises to verify and test the effectiveness of the software and the reliability of the algorithm,the test results show that the overall performance of the software is good,and the noise reduction algorithm is reliable and effective.(4)Taking the GNSS deformation monitoring data of Poyang Lake Second Bridge as an example,after comparing the three gross error detection algorithms,the GNSS bridge deformation monitoring data was preprocessed using the 3σ criterion.The improved LMD method is used on the GNSS＿DNR software to reduce the noise,and the fast Fourier Transform(FFT)is used to extract the vibration frequency of the bridge from the deformation sequence after the noise reduction,and analyzes the frequency spectrum of two GNSS monitoring stations located at symmetrical positions in the middle of the bridge.The results are highly consistent,indicating the effectiveness and practicability of the improved LMD noise reduction algorithm proposed in this paper for the identification of bridge structure modal parameters. |
