Construction Control Of Rigid Frame Bridge Parameter Identification Based On BP Neural Network

Long-span continuous rigid frame bridge with corrugated steel webs mostly adopts cantilever construction method.In the construction process,there are many design parameters that can affect the bridge structure and have great influence on the bridge structure.In the construction process,it is necessary to focus on it,correct the model parameter value in time,and reduce the construction deviation.When it is difficult to obtain the actual parameters of the structure directly by measuring,the parameter values can be calculated by parameter identification.Taking the auxiliary channel bridge of Yinzhouhu Bridge as the engineering background,the static analysis and parameter sensitivity analysis of corrugated steel web rigid frame bridge are carried out by Midas/Civil software,and the parameter identification of concrete bulk density and prestress loss is completed by Matlab software.The main work and achievements are as follows:(1)This paper introduces the engineering background of the auxiliary channel bridge of Yinzhouhu Bridge,establishes the full-bridge model of the bridge by using the finite element software Midas/Civil,and analyzes and calculates the stress and deformation of the rigid frame bridge in the maximum cantilever stage,the completion stage and the completion ten-year stage.The results show that from the maximum cantilever stage to the completion ten-year stage,the maximum compressive stress of the main girder roof and floor occurs near the main pier,and the maximum stress value meets the design specification requirements.(2)Parameter sensitivity analysis of continuous rigid frame bridge is carried out to analyze the influence of main beam weight,prestress loss and relative humidity on bridge structure.It is found that concrete bulk density and prestress load have great influence on the alignment of continuous rigid frame bridge,which is a sensitive design parameter,while environmental relative humidity has little influence on the alignment,which is a secondary design parameter.In the completed bridge state,the prestressed load has the greatest impact on the stress of the main beam,followed by the concrete bulk density,and the environmental relative humidity has the smallest impact.According to the results of parameter sensitivity analysis and based on the mid-term elevation data of main girder construction,the parameters of concrete bulk density and prestress loss are identified to guide the later construction.(3)The specific process of BP neural network algorithm applied to parameter identification of continuous rigid frame bridge with Matlab software is introduced.The trained BP neural network model has been tested and the accuracy fully meets the requirements of the bridge.Based on the elevation data of the main beam before and after concrete pouring and before and after tension,the concrete bulk density obtained by back analysis of BP neural network is 26.5228kN/m3,and the prestress loss is5.39%.The calculated value of the modified model is closer to the measured value than that of the original design,which verifies the feasibility of this method.