Optimization Of Lifting Cable Force And Linear Prediction Analysis Of Steel Box Arch Bridge

Steel arch bridges have been favored in the field of bridge construction in recent years because of the advantages of light weight,strong spanning ability,beautiful line shape,and material force calculation in accordance with calculated values.The arch rib construction of long-span arch bridges often adopts cable hoisting and oblique-stayed buckle hanging method.Due to the terrain restrictions,the S348 Shexian Xin’anjiang Bridge adopts a single-sided tower and the other side of the buckle is directly anchored on the mountain.For buckle hanging construction,The length of some buckles reaches about 300 m.Based on this project,this paper studies the optimization of buckle anchor cable force and the control of arch rib linearity,internal force and stress.The main work content is as follows:(1)The main non-linear influencing factors in the cable hoisting and oblique-stay buckle construction process of long-span steel box arch bridge are analyzed.For the long cable structure,taking its sag effect into account,a kind of spatial catenary clue unit is defined through the self-programming function of the Matrix27 unit of ANSYS.Based on the catenary clue element theory,a calculation program for solving the tangent stiffness matrix and cable end tension of a known unstressed cable length and a calculation program for solving the unstressed cable length for a known cable end tension are respectively given.The integral ANSYS finite element model of the oblique-stayed arch rib is established,and the coordinate update formula for the tangent assembly of the arch rib is derived,combined with the function of the life and death element,to simulate the tangent assembly process of the arch rib.(2)In order to avoid repeated cable adjustments during the construction process,the rigid support-elastic cable method and the backward analysis method are used to calculate the one-time tension cable force of the buckle cable.The results obtained basically meet the requirements,but the former has a large displacement of the arch ribs in the latter segments and the increase and decrease of the buckle force is uneven,and the latter obtains a larger initial tension cable force as the buckle is assembled in subsequent segments.Subsequently,based on the influence matrix method,the influence matrix of the arch rib control point displacement and the minimum bending energy on the buckle cable force were extracted respectively,and the cable force was optimized by the improved forward installation iteration method and the minimum bending strain energy method.After optimization,the vertical displacement,internal force and stress of the arch rib are greatly reduced,and some cable forces are more prominent.(3)In order to quickly solve the reasonable buckle cable force that satisfies multiple constraints,the optimization theory is introduced.Compare the optimization efficiency of SGD method,Momentum method and Ada Grad method in the first-order method.Finally,the Ada Grad method with a smoother convergence curve is selected as the optimizer.Establish an objective function for the sum of squares of vertical displacements of arch rib control points.Take the arch rib displacement,section stress,and buckle cable strain as constraints.The matlab first-order method optimization program is compiled,and the ANSYS forward analysis model is combined to solve the numerical solution of the global optimal cable force during the construction phase.The vertical displacement of the arch rib is less than ±2cm,the internal force and stress are optimized,the strain of the buckle is uniform,and the convergence speed is fast,and a good optimization effect is achieved.(4)During the hoisting process of the arch rib of the Xin’anjiang Bridge,a linear prediction model of the arch rib was established based on the least squares support vector machine method.Select the main parameters that affect the shape of the arch rib,analyze the effects of ? and ? on the model training fitting ability and test generalization ability,and determine their reasonable values.Train the model to predict the linear deviation during the construction of the second half of the arch rib.The results show that the training model based on the least square support vector machine method has a strong learning ability.The mean square errors of the training results of the side arch rib control points of the large and small mileage are 0.188 and 0.074,respectively,and the mean square errors of the prediction results are 0.312 and 0.358,respectively,and the error is not more than 1mm.The accuracy of the prediction results meets the engineering requirements and can be used as a basis for subsequent construction control,and provides a reference for error adjustment in the arch rib assembly process.