Abstract: The first outcrossing method can get the change of failure probability with time， but the calculation is complex and may call for more cost. The quasi-static method requires a smaller amount of calculation， but it can only get the failure probability value corresponding to a given period of time， and cannot obtain its change trend.In view of the above problems， AK-MCS-T， a time-dependent reliability analysis method is proposed by combining the Kriging model and the Monte Carlo method. It not only has the advantage of small calculation amount of the quasi-static method， but also can give the change of failure probability with time. AK-MCS-T proposes a new selection criterion for the Kriging model which not only considers the probability of wrong judgment of response evaluation value based on the Kriging model at a sample point， but also considers the probability density corresponding to this sample point. At the same time， a new learning stopping condition is proposed to ensure the accuracy of AK-MCS-T. Finally， the effectiveness and accuracy of the new method are proved by two examples: acorroded beam structure and a cantilever tube structure.

Key words: Monte Carlo method; Kriging model; selection criterion; learning stopping condition; time-dependent reliability analysis