公路工程

为实现大跨径斜拉桥有限元模型的合理精准修正，提出了一种基于双加点改进Kriging模型和非线性蜉蝣优化算法的有限元模型修正方法。首先采用U学习函数加点和考虑约束边界的加点准则构建了大跨径斜拉桥Kriging响应面模型，其次通过Logistic混沌映射和非线性重力系数改进了蜉蝣优化算法，随后以某大跨度斜拉桥荷载试验下主梁中跨跨中挠度最大工况和模态频率响应为目标函数建立了斜拉桥有限元参数修正模型，最后基于双加点Kriging和IMA算法实现了有限元参数的寻优求解。结果表明：采用双加点的改进Kriging模型测试集样本下平均相对误差仅为1.61%,可以较好地反映斜拉桥在不同效应下的响应特征；相较于同类型算法，IMA算法对有限元模型参数修正问题的收敛速度和精度更高；修正后三级静载工况下的主梁中跨跨中挠度误差均降低至3%以内，前四阶模态频率误差由14.9%降低至5.3%,验证了该修正方法的可行性。

To achieve rational and accurate correction of finite element models for long-span cable-stayed bridges, a finite element model correction method based on a double-add-point improved Kriging model and a nonlinear mayfly optimization algorithm is proposed.Firstly, a Kriging response surface model for long-span cable-stayed bridgesconstructed using a U-Learning function add-pointcriterion and anadd-point criterion considering constraint boundaries.Secondly, the mayfly optimization algorithm was improved through Logistic chaotic mapping and nonlinear gravity coefficients.Subsequently, using the maximum mid-span deflection case and modal frequency responses of the central span of the main girder under load test of a certain long-span cable-stayed bridge as objective functions, a parameter correction model for the cable-stayed bridge finite element model was established.Finally, based on the double-add-point Kriging and IMA algorithm, the optimization solution for finite element parameters was realized.The results show that: the improved Kriging model using double-add-point achieves an average relative error of only 1.61% on the test set samples, and can better reflect the response characteristics of cable-stayed bridges under different effects; compared to similar algorithm, the IMA algorithm exhibits higher convergence speed and accuracy for the finite element model parameter correction problem; after correction, the error in mid-span deflection of the central span of the main girder under three-level static load condition are all reduced to within 3%, and the error of the first four modal frequencies are reduced from 14.9% to 5.3%, verifying the feasibility of the correction method.