Research On Structural Damage Identification Based On Wavelet And Artificial Neural Network

In nearly ten years,the flow of ur ban and rural population has greatly increased.Existing roads,bridges and other construction has been unable to m eet the needs of the populatio n mobility.With the "13t h Five-Year" plan is put forward,all kinds of bridges,roads and other construc tion is in an unprecedented developm ent.At the same time,it has laid a solid founda tion for the transportation and econom ic growth of our country.However,the rapid development of the economy at the sam e time,is being built or have been built in the construction,design,quality and so on a series of construction,design,quality,and so on a series of security issues.The structure of the bridge,beam bridge is occupies a substantial proportion.The use of road transport and urba n transport is the most.Different forms of beams,as the main components of the beam bridge,are subjected to the combined effects of vehicle load,pedestrian load,wind load,wind and rain,and so on.Under the long-term action of these loads,the beam will suffer different degree of damage in the operation of the work,and will be ove r time,the dam age location and d egree of damage will have different degrees of developm ent.And wh en these dam age to a certain degree of development will lead to a substantial dec line in the stif fness of the beam body,and even on the entire bridge of the journey to bring security risks.Caused by the dam age of structural damage is often brittle failure,this destruction is no sign of dam age,will lead to serious acciden ts,causing serious social influence and is not possible to estimate the econom ic losses and casualties.Therefore,it is nece ssary to develop a complete health monitoring system in service bridge,tim ely monitoring of its operating conditions.And through a lar ge number of monitoring data to assess the health status of the b ridge,the most im portant work is to de termine the damage identification.In this paper,wavelet transform and neural network are used to diagnose the damage of a span sim ply supported beam.Its m ain principle is: F inite element calculation model of beam is established by using finite element calculation software MIDSA CIVIL.Based on the finite elem ent analysis of the dynamic characteristics of the beam,the modal parameters of the curvature are obtained.Then,using the wavelet and neural network toolbox of MA TLAB7.0,the modal parameters of the beam are analyzed.Finally,determine the location and degree of dam age identification.The contents and conclusions of the study are as follows:(1)In this paper,based on the curvature mode,the wavelet transform can be used to amplify and identify the abrupt change signa l in the structural response at dif ferent scales,A method of structural dam age identification based on wavelet transform is proposed.The single dam age and multi dam age damage simulation of a structu re numerical model is carried out by using the method of unit stiffness reduction.Then,the natural frequency of structure in dam age state is obtained by using the finite element software MIDSA CIVIL,and the corresponding curvature modes are obtained by the dif ference method.Finally,the wavele t coefficients of the curvature mode difference of structure is obtained by using wavelet analysis.By identifying the abrupt point of the wavelet co efficient curve,th e damage position and dam age degree of structure is determined.(2)Artificial setting up 5 kinds of da mage cases,and to collect th e first 5 frequency mode changed with dif ferent injury conditions to establish the BP neural network training samples,and on their training.Then,the unknown damage conditions on beam,damage identification using BP neural network training after optim ization,that is able to locate dam age well location.Finally,for identification of injury degree also establish five frequencies of modal pa rameters of the sam ple,establish for BP neural networks to identify the dam age extent,set the stochastic dam age parameters,the use of training the optimized BP neural network on the beam damage identification,found recognition errors are controlled within the range of 5%.It proves that BP neural network for damage location and damage extent identification can be used.