论文部分内容阅读
服役多年的桥梁由于受环境和车辆的影响,依据设计资料建立的有限元模型已不能准确地反映桥梁的实际受力性能。该文以某桥为背景,对该桥2片服役20年的PC空心板梁进行破坏试验以及有限元建模,利用BP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PC空心板梁仍具有良好的刚度与弹性恢复能力;修正后的有限元模型与实际结构的物理状态非常接近,挠度误差均在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。
Due to the influence of environment and vehicles, the finite element model based on design data can not accurately reflect the actual mechanical properties of bridges. In this paper, a bridge is taken as the background, two 20-year-old PC hollow slab beams are tested for failure and finite element modeling, and the BP neural network is used to correct the finite element model. First of all, taking the mid-span displacement of the finite element model under different structural parameters as input, the corresponding structural parameters such as stirrup, longitudinal reinforcement, strand, concrete elastic modulus and Poisson’s ratio of concrete are taken as output to calculate the finite element Model’s design parameters. The results show that the 20-year PC hollow slab beam still has good stiffness and elastic recovery ability. The modified finite element model is very close to the physical structure of the actual structure, and the deflection errors are all within 5%. The modified strand The modulus of elasticity is in good agreement with the experimental value, which proves the accuracy and rationality of the correction results.