Research On Seismic Vulnerability Analysis Method Of Bridges Based On Influence Line Model Modification

In order to meet the technical requirements of bridge structures with multi-level target seismic resistance,the ideas concerning performance-based seismic protection of bridges has gradually become a hot topic of interest for scholars at home and abroad,among which seismic susceptibility analysis,as a key link in the ideas concerning performance-based seismic protection,boasts a significant application value and research potential,and its research work has made considerable progress.However,there still remain some unavoidable problems in the seismic susceptibility analysis of existing bridge structures: 1.the uncertainty of the finite element model of existing bridge structures is often ignored and the influence of finite element model correction on the analysis results has not been considered when the susceptibility analysis is implemented in existing bridges;2.the traditional bridge structure model correction research based on dynamic method is vulnerable to the influence of test noise in the actual measurement environment,and meanwhile the study of quasi-static influence line model correction method on the basis of neural network for bridges is still at its initial stage.Based on the above-mentioned scientific problems,the bridge influence line model correction study on the basis of BP neural network has been carried out,and the seismic vulnerability analysis method of existing bridges considering the model correction has been further investigated.Focusing on the seismic susceptibility of existing bridges,this paper implements bridge structure model modification research on the initial bridge model built in accordance with engineering drawings based on BP neural network algorithm with considering the uncertainty of the bridge structural model cross-sectional parameters,and further proposes the seismic vulnerability analysis technical route on the basis of the model correction and validates its feasibility.Considering such problems as environmental noise in dynamic test data and insufficient sample size of static test data,the method of constructing the objective function by using quasi-static strain influence line data is carried out the study of model modification to conduct the model modification research based on strain influence line of bridge.In accordance with the Beta probability density,the bridge exceedance probability is fitted to convert the structural susceptibility matrix into a seismic damage index-seismic intensity measure expression,which comprehensively evaluates the seismic performance of existing bridges.Based on the above-mentioned research,the main achievements are as follow:(1)This paper reviews the current status of research on finite element model modification and seismic vulnerability analysis of bridges,discusses the deficiency of the traditional model correction in constructing the objective function with dynamic and static force test data,and proposes the method of model correction using quasi-static influence line data as the objective parameter;on the basis of pointing out the limitations of building finite element models only according to engineering design drawings and of carrying out seismic evaluation,this paper puts forward the method of seismic vulnerability analysis based on model correction to further improve the scientific and validity of seismic evaluation results of existing bridges.(2)Without considering the influence of model correction theory on the numerical simulation results,taken a three-span steel plate composite girder bridge as the test sample,this paper selects 20 ground motion records and 1 artificial seismic wave according to the normative response spectrum and spectral coincidence criterion of the area where the test bridge is located,and also conducts the moment-curvature analysis on the pier column section of the test bridge on the XTRACT analysis platform,and thus this paper defines the damage index threshold of each performance level,calculates the probability of exceedance at different performance levels and seismic intensities based on the incremental dynamic analysis method,and further convert the damage susceptibility matrix into a seismic damage index expression based on the Beta probability density distribution to carry out the seismic damage index analysis on the initial model of the test bridge.(3)On basis of the theory of finite element model modification,with the limitations of constructing the objective function with dynamic and static load data being considered,the quasi-static strain influence lines of bridge are extracted from the pseudo-static test of an actual existing bridge to build BP neural network based on the MATLAB operating environment and BP neural network would do self-learn through data sample.And thus this paper proposes a model modification method based on strain influence line with the combination of BP neural network is proposed to conduct finite element model correction test research on bridges.The proposed objective function based on the strain-influence line can more comprehensively evaluate the model deviation between corrections;the proposed model correction method can update and adjust the correction parameters scientifically and efficiently,and the modified iterative model has higher fidelity than the initial model.(4)Considering the randomness and uncertainty in the modelling process,this paper conducts the analysis of the seismic susceptibility after modification through introducing model modification theory and obtains the conditional probability of the structure exceeding each performance level.In addition,in order to accurately and quantitatively analyze the effect of model modification on the existing bridge model modification,this paper proposes a seismic damage optimization ratio curve based on the Beta probability density distribution theory to further quantify the vulnerability of the test bridge between model modifications.The results show that the model uncertainty has an increasing influence on the seismic susceptibility of bridges as the intensity of ground motion increases;the proposed Optimization Ratio Curve of Seismic Damage Index can visually reflect the degree of bridge model optimization,and the Beta probability density curve can provide new ideas for predicting the seismic performance of bridge lifeline facilities under severe earthquakes.Figure [41] table [24] reference [94]...