Abstract: To study the chief factors that affect the precison of disk LCF life estimation models, and improve evaluate precision of the models, the material parameters of three strain-life estimation models were obtained by heteroscedastic regression analysis method, and the fatigue life distribution of a aeroengine fan disk was simulated. The probabilistic life of the disk was finally obtained. Compaired with the experimental data, LCF life estimated by the heteroscedastic regression method is better than one by the homoscedastic regression method. The LCF life predicted by the equivalent strain model is much higher than the experimental value,while the results of the first principal strain model fit well. The estimated value of the maximum shear srain model is between the estimated value of other two models. If average extending stress was replaced by peak extending stress for the maximum shear srain model, the predicted probabilistic fatigue life agrees well with the experimental data. 

Key words: Low cyc fatigue; Heteroscedastic regression; Probabilistic life; Local strain-life method

摘要： 为了研究影响轮盘结构寿命预估精度的主要因素,提高模型预测精度,通过对三种材料应变概率寿命模型参数的异方差回归,对某航空发动机风扇盘榫槽部位疲劳寿命分布进行了数值模拟,得到了风扇盘榫槽部位的概率寿命。经与试验结果对比表明,考虑材料疲劳寿命异方差特性比按同方差处理更为合理。等效应变模型对疲劳寿命严重高估,第一主应变模型预测值与试验结果吻合较好。最大剪应变模型预测值介于以上两个模型之间,用峰值拉伸应力代替平均拉伸应力进行修正后最大剪应变模型预测值与试验值吻合良好。 

关键词: 低周疲劳;异方差回归;概率寿命;局部应变-寿命法