ABSTRACT
To build a reasonable resistance model, the inverse Gaussian (IG) model is employed to describe the deterioration process in this paper. A bridge structure time-dependent reliability model is proposed based on the IG stochastic process model. The expectation maximization (EM) algorithm estimates three parameters according to the resistance deterministic function. The results show that the IG stochastic process model is superior to the deterministic function in the time-dependent reliability model. When the base period T = 30a, the reliability of the bridge under the annual load growth rate of 1.0 and 2.0 is 0.797 and 0.6512, which are 0.897 and 0.733 times that of the stable load (q = 0), respectively. The degradation rate parameter β in the IG model is higher sensitive than the parameter η for the structural reliability probability. More attention should be paid to the accurate value of the degradation rate β when the IG process model is employed.
