摘要：Uncertainties in material properties, geometricdimensions, loads and other parameters are always un-avoidable in engineering structural problems [1-4]. Theyhave played a more and more important role in the struc-tural reliability analysis. In order to obtain the objective ofreliable design, the effects of the various uncertain parame-ters should be rationally considered and treated. The pro-babilistic models are widely used to describe the uncertain-ties and they have been proved very effective in the struc-tural reliability problems [5-7]. However, it is difficult toestimate precise values of parameters to accurately definethe probability distributions because of inaccurate and in-sufficient information. Once the assumption about theprobability distributions is not satisfied, the structural reli-ability analysis seems doubtful and meaningless. More-over, some researches [8-11] have also indicated that evensmall deviations from the real probability distributions maycause large errors in the reliability analysis.
