样本容量和分布类型对疲劳寿命分布的推断影响分析

Analysis of the effects of sample size and distribution model on the inference of fatigue life distribution

应用蒙特卡罗数值仿真方法,就样本容量及分布类型的选取对产品疲劳寿命概率分布函数推断的影响进行了研究,给出不同样本容量及分布函数类型下,各分布函数通过拟合优度检验被接受作为产品疲劳寿命概率分布函数的概率。还探讨了选取非真实分布函数作为产品疲劳寿命概率分布函数对累积失效概率的影响,并将对应给定百分位点下,由非真实分布函数求得的累积失效概率与真实累积失效概率进行了对比分析。所得结论对产品的设计和安全性评估均有参考价值。

Using Monte Carlo simulation method,the effects of sample size and distribution model on the inference of the fatigue life PDF are investigated. The probabilities of each distribution function accepted by goodness-of-fit as the product＇s PDF for different sample sizes and different distribution models are provided. Moreover,the effects of choosing unreal distribution funetion as the product＇s PDF on the Cumulative Failure Probability are discussed,and the cumulative failure probability corresponding a specify percentile,which is computed by unreal distribution,is compared with its true value .The conclusion may be useful to the design and the safety evaluation of product.