The Structural Identification Of Small And Mudium Bridges Based On Big Data Of Vehicles And Interval Inverse Analysis With Affine Arithmetic

The increasing traffic flow and common overloading speed up the bridge deterioration process and cause the bridge collapse frequently,how to evaluate the bearing capacity,operation state and durability of small and medium-sized bridges in service reliably and quickly under the premise of reducing cost investment has always been an urgent problem to be solved in the field of bridge health monitoring.As an important characteristic curve of bridge structure,the influence line(IL)or influence surface(IS)is capable of conducting professional assessment on the stiffness distribution,damage condition and bearing capacity of the existing bridges.However,the commonly used measurement methods of IL have deficiencies such as taking time and effort and hindering the traffic in practical applications.Therefore,this paper proposes a non-contact bridge structural identification method based on big data of vehicle excitations and interval analysis with affine arithmetic in view of the development and urgent needs of small and medium bridges monitoring techniques.This paper mainly studies the following aspects,(1)In theory,this paper combines the IL calculation method and interval analysis with affine arithmetic to establish the inversion model to directly calculate the interval solution of the bridge IL from the time-history responses when the axle weight parameters are uncertain interval variables.The support vector machine(SVM)is used to extract the deterministic IL or IS from big data of interval solutions under multiple working conditions.Finally,the equation of typical bridge weigh-in-motion(BWIM)based on the calibrated IL is derived,and then the compound feedback verification mechanism based on the mutual deduction of axle weight intervals and IL interval under multiple working conditions is proposed to correct the initial over estimated axle weight intervals,so as to improve the identification accuracy of IL.(2)In the aspect of numerical simulation,this paper establishes the mapping relationship between vehicle type and axle load interval by referring to the research of parameter statistics and classification of two axle vehicles in actual traffic.The finite element model of beam-slab bridge based on ANSYS and the numerical model of spring damping half car based on MATLAB are established respectively to simulate the IL intervals of bridge corresponding to vehicle types and axle-weight intervals under various cases of load,speed and vehicle type.The IL intervals under multiple testing cases are separated into boundary points,and the binary classification algorithm of SVM is used to identify the deterministic displacement IL in the two-dimensional plane.At the same time,based on the idea of closed-loop verification,the SVM identified results are employed to back-calculate the moving axle loads for a realization of contactless bridge weigh-in-motion(c BWIM).Then,several examples are used to verify the effect of axle loads correction and the improvement of IL identification accuracy by compound feedback verification mechanism between the axle load range and the IL range.(3)In the aspect of test,the scaled models including a simply supported T-beam bridge and a three-axle truck are used to carry out the simplified vehicle-bridge coupling vibration test,laser displacement meter and strain gauge are used to measure the structural dynamic responses,which are employed to verify the proposed IS identification method utilizing vehicle big data and interval analysis with affine arithmetic.The IS of strain and displacement of the two-lane model bridge is calibrated by dynamic test,and the error of IS identified by SVM algorithm are verified to be acceptable.At last,the identified ISs are employed for weighing vehicle loads to realize the c BWIM,which further proves the feasibility and accuracy of the non-contact IS identification method proposed in this paper.