论文部分内容阅读
Firstly,the fatigue damages associated with the random loadings were always deemed as highcycle or very-high-cycle fatigue problems,and based on Chebyshev theorem,the number of rainflow cycles in a given time interval could be recognized as a constant by neglecting its randomness.Secondly,the randomness of fatigue damage induced by the distribution of rainflow cycles was analyzed.According to central limit theorem,the fatigue damage could be assumed to follow Gaussian distribution,and the statistical parameters:mean and variance,were derived from Dirlik’s solution.Finally,the proposed method was used to a simulate Gaussian random loading and the measured random loading from an aircraft.Comparisons with observed results were carried out extensively.In the first example,the relative errors of the proposed method are 2.29%,3.52%and 1.16%for the mean,standard deviation and variation coefficient of fatigue damage,respectively.In the second example,these relative errors are 11.70%,173.32%and 18.20%,and the larger errors are attributable to non-stationary state of the measured loading to some extent.
Firstly, the fatigue damages with the random loadings were always deemed as high cycle or very-high-cycle fatigue problems, and based on Chebyshev theorem, the number of rainflow cycles in a given time interval could be recognized as a constant by neglecting its randomness .Secondly, the randomness of fatigue damage induced by the distribution of rainflow cycles was analyzed. According to central limit theorem, the fatigue damage could be assumed to follow Gaussian distribution, and the statistical parameters: mean and variance, were derived from Dirlik’s solution. Finally, the proposed method was used to simulate Gaussian random loading and the measured random loading from an aircraft. Comparisons with observed results were carried out extensively. In the first example, the relative errors of the proposed method are 2.29%, 3.52% and 1.16% for the mean, standard deviation and variation coefficient of fatigue damage, respectively. In the second example, these relative errors are 11.70%, 173.32% and 18.20%, and the larger errors are attributable to non-stationary state of the measured loading to some extent.