构建了耐热钢的寿命预测及可靠性分析非参数统计模型,较已有的统计模型相比,有很好的稳健性和较高的精度.首先利用局部线性回归得出已提出的双指数主曲线.其次为了可靠性分析,对所得Z参数进行正态性检验.当Z参数正态性检验不显著时,利用核密度估计Z的分布密度,建立Z参数与可靠度的关系,得出应力-P参数-可靠度曲线.最后以2.25Cr-1Mo 耐热钢蠕变断裂数据为例建立相应的统计模型并检验模型分析效果.%The nonparametric statistical model is established on life prediction and reliability analysis of heat-resistant steel.Compared with existing models,it has good robustness and better precision.Firstly,the bi-exponent shape master curve is obtained by local linear regression.Secondly,In order to have a reliability analysis,we have a test of normality on Z parameters in Shapiro-Wilk test,the result of test shows that the normality assumption on Z parameter is not significant.The kernel density estimation is used to obtain the distribution density of Z parameter,The relationship between Z parameters and reliability is established,the predicting curves of σ-P-Rare constructed.Finally,the reliability analysis of 2.25Cr-1Mo,as example,is used to test on the effect of models.
展开▼