The Research On Automatic Identification Method Of Mode Parameter For Bridge Structure Under Ambient Excitation

Bridge structure modal parameter identification has great significance in the damage identification of bridge health monitoring system as well as the performance evaluation.Though many scholars,at home or abroad,have achieved remarkable progress in the signal de-noising and modal parameter identification by means of ensemble empirical mode decomposition and the stochastic subspace algorithm,there still exist some shortcomings on the poor adaptability and low-level automation and so on.This paper carries out in-depth theoretical and testing research in regard to the deficiency of EEMD and SSI,relying on “the method study on behavioral evolution theory of bridge structure and its safety monitoring”,which is a part of National Basic Research Program of China(2012CB723305),undertook by the professor.The main study process and production are as follows:(1)In view of the insufficient of the ensemble empirical mode decomposition(EEMD),an improved signal decomposition algorithm named AEEMD is proposed,The main contents include: i)deduced a theory formula considering standard deviation of added white noise amplitude and the average integration times,according to the added rule of white noise and the optimal relationship between white noise amplitude standard deviation and the amplitude standard deviation forhigh-frequency component of response signal;ii)proposeda new method of endpoint effect-polygon positioning continuation algorithm,taking advantages of the predictprolongation algorithm and extreme value point prolongationalgorithm,to solve the problem that bridge structure with lower natural frequencies and high sensitive degree of end effects;iii)introduced de-correlation algorithm and clustering analysis algorithm in the process of responsesignaldecomposition,in reaction to the problem of modal-aliasing in IMF component;iv)put forward an effective filter coefficient of intrinsic mode function that is effective information coefficient,by using the linear weighted average algorithm with the correlation degree,the information entropy,energy density and its average period to overcome the problems needed to be artificial screening;v)written the procedure code of AEEMD by Matlab,and conducted feasibility analysis and verification of the proposed algorithm by simulation signal and real response signal of bridge structure.(2)In this part,modal parameters automatic identificationfor bridge structure(ADATA-SSI)are proposed by aiming at the limitation of the data-driven stochastic subspace algorithm.The main contents include: i)proposed a novel singular entropy incremental derivative method,based on the existing order determining method of singular value jumping and stabilization diagram,the error from the man power is avoided.in view of the difficulty in order determining which needs human power.ii)introduced “the sliding window” to solve the problem of the data-driven stochastic subspace algorithm with low reliabilityin parameter identificationfor structure under the time-varying condition.Based on this,a system in the whole window can be regarded as a time invariant system;iii)put forward a real modal filter algorithm which regards frequency as the main cluster basis,damping ratio and mode as theverifiable basis,considering thelimitation of human participation in real modal filter for bridge structure;iv)came up with an improved automatic identification of mode parameter for bridge structureby combining the vibration signal preprocessing,AEEMD algorithm and ADATA SSI algorithm,thenwrittenits procedure code by Matlab.(3)Using the Fuling Yangtze river bridge test data,the signal noise reduction and reconstruction are implemented on the main girder of this bridge by using AEEMD algorithm and EEMD algorithm respectively,to verify the proposed method AEEMD has a better effect of decomposition and reconstruction;Moreover,the proposed method is implemented to identify the mode parameter of response signal which is reconstructed automatically,and then the comparative analysis among the identified results,experimental value and FEM results are implemented to verify the feasibility and superiority of the proposed method.(4)By making use of the monitoring datas in the operation phase of the Sutong Bridge,firstly,make sure that whether the collecting signal of sensor by using exploratory data analysis method is available;And then the vibration signal is smoothed by adopting the five-spot triple smoothing algorithm;Finally,using the proposed method which is based on the AEEMD and ADATA-SSI to identify the modal parameters of the response signal in 2015 for Sutong Bridge,and a comparative analysis is implemented on its final identified results and the theoretical value from different literatures,at last,the condition of mode parameter with the change of time is also analyzed by taking a day as the smallest unit.