Abstract: In view of the fact that the failure state of the performance function in the traditional design is often inconsistent with common sense, a reliability model of fatigue life which takes into consideration the fuzzy failure criteria is presented. The fatigue life and the fatigue design life are treated respectively as a random variable and a fuzzy random variable, and reliability is analyzed when the fatigue design life obeys linear L-R distribution. The probability density function of fatigue life under random load is deduced. A new numerical algorithm for fuzzy reliability analysis with random variables and fuzzy variables is developed. The double integration to calculate fuzzy reliability is first reduced to univariate integration by use of Gauss-Legendre integration. Thresholds corresponding to Gauss-Legendre abscissas are listed, and the fatigue design life is truncated to the subset under the given threshold. Then the probability density function of fatigue life is supposed to be of uniform distribution in the truncated subset. The reliability of structure fatigue life corresponding to the threshold can be obtained using Monte Carlo simulation. Numerical results show that the fuzzy reliability analysis of fatigue life tends to provide more safety than traditional analysis does.

Key words: fuzzy failure criterion, fatigue, reliability, Gauss-Legendre integration, Monte Carlo simulation