基于可靠度的桥梁船撞作用荷载组合分项系数
Combined partial coefficient of ship？？bridge？？collision loads based on reliability

对船撞偶然组合问题，现行的桥梁设计规范给出了相应的组合分项系数。为了分析现行规范给出的分项系数下桥梁结构可靠度指标水平，寻求荷载组合分项系数与桥梁结构可靠度指标之间的关系，需要对桥梁船撞偶然荷载组合进行深入研究。基于有限元？采窬？网络？？Monte Carlo(F？？A？？M)法和极限状态设计法，计算桥梁结构船撞偶然组合不同荷载分项系数下的可靠度指标。以练江1号桥主桥为依托工程，采用有限元软件建立全桥空间有限元动力模型，以荷载冲击谱模拟船撞桥墩的动力时程，塔克斯特拉准则作为荷载组合理论依据，确定船撞荷载与汽车荷载作用下结构的失效模式。提取船撞荷载与荷载效应样本、汽车荷载与荷载效应样本、船撞和汽车荷载组合与荷载效应样本，对这些样本进行BP神经网络训练，当训练结果满足精度要求时，对船撞荷载分项系数？？ψ？？？？C？？？？V？？取1.0，汽车荷载分项系数？？ψ？？？？L？？？？L？？分别取0.5、0.6、0.7、0.8、0.9、1.0，并采用Monte Carlo法计算各分项系数下桥梁结构的失效概率和可靠度指标，将不同分项系数下的可靠度指标与目标可靠度指标进行比较。研究结果表明：跨径在150 m以内的连续刚构桥荷载组合分项系数？？ψ？？？？C？？？？V？？、？？ψ？？？？L？？？？L？？，建议其值为1.0、0.8；F？？A？？M法可以方便、快速地求解桥梁结构船撞偶然组合下的可靠度指标，〖JP2〗建立可靠度指标？？β？？与船撞荷载分项系数？？ψ？？？？C？？？？V？？、汽车荷载分项系数？？ψ？？？？L？？？？L？？的影响面关系，并给出荷载分项系数建议值，为船撞偶然组合设计及桥梁船撞的风险评估提供依据。
For accidental ship collisions, the existing bridge design code has used the corresponding partial coefficient of combination. In order to analyze the reliability index level of the bridge structure under the partial coefficient given by the current specification, and find the relationship between load combination partial coefficient and bridge structural reliability index, it was necessary to study the accidental load combination of bridge？？ship collision in depth. Based on a finite element？？neural network？？Monte Carlo method and limit state design method, the reliability index of a bridge structure under different load partial coefficients was calculated. Taking the No.1 Lianjiang Bridge as the main bridge supporting project, finite element software was used to establish the full？？bridge space finite element dynamic model, and the load shock spectrum was used to simulate the dynamic time history of a ship colliding with a pier. Turkstra？？s principle was served as a theoretical basis for the load combination to determine the failure mode of the structure under ship impact and vehicle loads. The load and load effect samples of the ship collision, the vehicle, and the combination of ship collision and vehicle were extracted, and the BP neural network was trained on these samples. When the training results met the accuracy requirements, the partial coefficient ？？ψ？？？？C？？？？V？？ of the ship impact load was taken as 1.0, and the partial factor ？？ψ？？？？L？？？？L？？ of the vehicle load was taken as 0.5, 0.6, 0.7, 0.8, 0.9, and 1.0. The Monte Carlo method was used to calculate the failure probability and reliability index of the bridge structure under different partial coefficients. the reliability index under different partial coefficients and the target reliability index were compared. The results show that a ship collision load partial coefficient is 1.0 and a