| The seismic response model of the bridge structure can not only be used to identify and judge the damage of the bridge under earthquake,but also provide new ideas for the seismic design of the bridge.Therefore,it is of great practical significance to study the seismic response modeling method of the bridge structure.Traditional seismic response analysis methods include static analysis and time history analysis.Because the former ignores the higher-order modes,the results have large errors;the latter is difficult to be popularized due to the difficulty of parameter calculation.With the development of bridge health monitoring,a large amount of monitoring data is collected,which meets the data conditions of the neural network algorithm,which is an adaptive highly nonlinear modeling method.Therefore,this paper introduces the LSTM network in the neural netw ork algorithm to model the seismic response of a cable-stayed bridge.Then,a damage identification method based on the seismic response model of the bridge structure is proposed,and the model built in this paper is applied to the field of damage identification.The main research is as follows:1.Collect ground motion and bridge response data and establish a database.Firstly,data selection rules are proposed based on the characteristics of bridge response data;then the ground motion and bridge respons e time history data are collected according to this rule,and the collected data is preprocessed and statistically analyzed;the processed data is made into a database for subsequent neural Network modeling.2.The LSTM network is used to model the seismic response of the bridge structure.First,the basic mathematical principles and construction steps of the LSTM network are introduced;then the feasibility of the method for modeling the seismic response of bridge structures based on the LSTM network is an alyzed;then the calculation diagram of the LSTM network is designed,and the ground motion-pylon response model is established based on the calculation diagram And ground motion-box girder response model;Finally,the source of the error of the model results is discussed through correlation analysis and coherence function analysis,which provides ideas for future research.3.The bridge structure seismic response model established in this paper is applied to the field of damage identification.First,the K L divergence is used as the damage index to establish the reference range for damage discrimination.Then,the KL divergence value of the model predicted response and the initial response is calculated,and the calculated KL divergence value is compared wi th the reference range to determine the bridge’s health status.Verification of the proposed method for damage identification based on modal analysis. |