基于《公路钢管混凝土拱桥设计规范》和《钢管混凝土拱桥技术规范》系统的构建了大跨度钢管混凝土拱桥正常使用可靠度评估模型.针对钢管混凝土拱桥功能函数隐式化及高度非线性的问题,将BP神经网络、均匀设计法及改进JC法综合运用,提出了适用于这类复杂结构可靠度分析的混合算法,通过数值算例和实际结构验证了该算法的可行性和准确性.实桥工程评估表明:在正常使用极限状态下,大跨度钢管混凝土拱桥各构件可靠度水平差异较大,主拱应力和挠度可靠指标均远超规范的规定,具有较大的优化空间;横梁和立柱应力可靠指标偏低,应引起设计的注意.
Based on the codes entitled separately "Specifications for Design of Highway CFST Arch Bridges" and "Technical Code for CFST Arch Bridges", the reliability evaluation model is presented for long-span CFST arch bridges at serviceability limit state. Aiming at the implicit and nonlinear characteristics of CFST arch bridges, a hybrid algorithm for calculating the reliability index is proposed with combination the BP neural network, UD design method and JC method comprehensively. The feasibility and accuracy of the hybrid algorithm are proved through the numerical and structural examples. The analysis result of practical engineering indicates that, there is a large difference in the reliability levels of different components for long-span CFST arch bridges during the serviceability limit state. The arch rib can be further optimized due to its reliability indexes of deflection and stress highly exceeding the requirements of codes. While the reliability indexes of stress for cross beam and column of are at lower level, which should be concentrated by designers.