基于参数回归模型的2.25Cr-1Mo钢蠕变断裂时间统计分析  

Statistical Analysis on Creep Rupture Life of 2.25Cr-1Mo Steel on the Basis of Parameter Regression Models

作　　者：郑开云[1] 吾之英[1] 

机构地区：[1]上海发电设备成套设计研究院,上海200240

出　　处：《动力工程学报》2013年第12期980-988,共9页Journal of Chinese Society of Power Engineering

摘　　要：基于多种寿命数据分布,包括指数(Exponential)、对数正态(Log-Normal)、威布尔(Weibull)、广义Gamma以及对数罗吉斯提克(Log-Logistic)分布,采用Manson-Haferd时间-温度参数模型对2.25Cr-1Mo(T22)钢的蠕变断裂时间进行了回归分析,对比研究了基于不同分布模型的蠕变断裂时间与温度和应力的关系、模型拟合优度以及各种模型蠕变断裂时间中位值与5%分位值的差异.结果表明:除Exponential分布模型不适用以外,Log-Normal、Weibull、广义Gamma和Log-Logistic分布模型均适用于蠕变断裂时间数据;这4种模型对蠕变断裂时间中位值的估计均相近,但是对低分位值的估计有所不同,特别是在数据分散度较大时,蠕变断裂时间低分位值的差异较为显著,其中Weibull分布模型的蠕变断裂时间低分位值显著低于其他模型.因此,宜采用多种分布模型对2.25Cr-1Mo钢进行综合统计分析,以正确预测其蠕变寿命或长时服役应力.Creep rupture lives of 2.25Cr 1Mo （T22） steel were statistically analyzed, through Manson-Ha- ferd time temperature parameter method, on the basis of Log Normal, Weibull, Exponential, Gamma and l.og Logistic distribution parameter regression models, and subsequenty a comparison was made to the de- pendence of creep rupture time on the temperature and stress, the goodness of fit, and the median and 5th percentile of creep rupture time among selected models. Results show that all the distribution models ex- cept Exponential distribution are suitable for prediction of the creep rupture time, and the predited medians with the 4 models are very close, but their predictions on lower percentiles are different, expecially on scattered data, and the lower percentiles predicted with Weibull model are significantly lower than those with other models. Therefore, the creep rupture lifetime and long term stress of 2.25Cr-lMo steel should be comprehensively analyzed based on multiple distribution models.

关 键 词：2.25Cr-1Mo(T22)钢 蠕变断裂时间 回归分析 参数模型 极大似然估计 

分 类 号：TG135[一般工业技术—材料科学与工程]