Research On Damage Identification Of Girder Bridge Based On Wavelet Packet Energy Method

Any building structures,including bridge,which occupy the important position in the modern society traffic system.If the damage of bridge cannot be found out timely,may cause serious accidents and bring great threats to people lives and property security,even to social development and stability.So,the damage identification of bridges,especially for its basic form,simply supported bridge,has important significance for bridge maintenance,strengthening even reconstruction.Begin with two main aspects,wavelet packet analysis and build BP neural network,study damage detection of bridge from structure energy.The research contents:(1)Take the damage identification of a beam under moving loads to simulate the damage detection of actual bridge under the vehicle,and build the finite element simulation model of vehicle-bridge interaction.(2)The simulation analysis is carried out in the hypothesis of various damage conditions that may occur in practice.Make wavelet packet decomposition of the displacement response of the beam structure,and after the decomposition take the energy curvature difference of each frequency component as damage identification index,the numerical calculation is done and good results are obtained.(3)With the development of big data and smart cities,in order to facilitate the intelligent analysis of the damage detection combined with the neural network,normalized the energy curvature difference.Build the BP neural network model used in this paper which is used for damage detection,the results of the energy curvature difference analysis were verified,furthermore,the predict results of damage conditions are satisfactory.(4)This paper discusses several typical simulation conditions,the effect of noise is analyzed and the robustness of the method is good.At the same time,the neural network is optimized through genetic algorithm,the effectiveness and practicability of wavelet packet energy curvature difference method and neural network method are verified,the prediction of damage is further discussed.This paper applies the theory of wavelet packet analysis,combined with the neural network method,analyzes various typical operating conditions.As much as possible,consider the combination of theory and practice,studies the problem of the damage identification of the beam under moving loads.The results can provide some theoretical guidance for the engineering practice of intelligent damage diagnosis.