Because there're no the finite element models under initial conditions of old bridges and no long-term health monitoring data,it is a hot topic to simulate and analyze old bridges by using finite element updating meth-od. On the background of the failure test of a prestressed concrete hollow slab beam in service for 20 years and the finite element models of ANSYS,the models were modified by using MATLAB radial basis function neural network. After taking the mid-span displacements of the finite element models with different parameters as the input vectors and some structure parameters,such as the corresponding elastic modulus of stirrups, longitudinal reinforcement, steel wires,concrete and poisson ratio of concrete as the output vectors,the design parameters of the finite element models were calculated. The results show that the prestressed concrete hollow slab beam in service in for 20 years still has good stiffness and elastic recovery ability; the based on radial basis function neural network is taken to modify the finite element model and the actual structure of the physical state very close to the errors which are with-in 5%;modified steel wire elastic modulus and test value in good agreement to prove the rationality and accuracy of the results of modified.%服役多年的桥梁由于没有初始条件的有限元模型,也没有长期健康监测数据,因此通过有限元修正来对旧桥进行模拟分析一直是一个热点.基于某运营20年的预应力混凝土空心板梁的破坏试验以及ANSYS有限元模型,利用MATLAB径向基神经网络对有限元模型进行修正.首先以不同的结构参数条件下有限元模型跨中位移作为输入,以对应的箍筋、纵筋、钢绞线、混凝土的弹性模量及混凝土泊松比等结构参数作为输出,计算出有限元模型的设计参数.研究表明:服役20年的预应力混凝土空心板梁仍具有良好的刚度和弹性恢复能力;修正后的有限元模型与实际结构的物理状态非常接近,误差均在5%以内;修正后的钢绞线弹性模量与试验值吻合良好,证明修正结果的准确性和合理性.
展开▼
