Abstract:

:On the basis of the time-dependent reliability theory and by considering the three-grade working mode of bridges, a new method to estimate the risk probability of existing bridges, marked FNM （Finite Element-Neural Network-Monte-Carlo Simulation） , is proposed. In this method, the time-dependent factors of both load and resist- ance are considered, some random samples of risk factors at a certain time in the operation cycle are obtained via the Monte-Carlo method based on the risk identification, and the risk probability is calculated by employing the neural network and the finite element method. Moreover, a case study of an 80 m-span concrete arch bridge that has been in operation for 8 years is performed. The results indicate that the proposed FNM method is effective in estimating the bearing capacity and risk probability at a certain time in the operation cycle. The proposed FNM method can also provide references to the decision-making of reinforcement of existing bridges.

Key words: existing bridge, three-grade working mode, Monte-Carlo method, time-dependent reliability, risk probability