江苏省科技期刊可视化阅读平台

导航条

江苏科技信息 ›› 2016, Vol. 33 ›› Issue (15): 59-62.doi: 10.3969/j.issn.1004-7530.2016.15.024

• 论文 • 上一篇    下一篇

疲劳寿命分布参数估计方法对比

奚蔚, 黄甫, 赵剑军   

  1. 上海飞机设计研究院强度设计研究部,上海,201210
  • 出版日期:2016-05-25 发布日期:2016-05-25

Comparison of Parameter Estimation Methods for Fatigue Life Distribution

Xi Wei, Huang Fu, Zhao Jianjun   

  • Online:2016-05-25 Published:2016-05-25

摘要： 疲劳寿命分布通常用威布尔分布描述，文章介绍了威布尔分布参数估计的3种方法，极大似然法、最小二乘法和头两序数法。用它们对不同抽样次数和样本容量下的样本进行了参数估计，并对比了估计结果。对比表明，当抽样次数很小时，宜用极大似然法和最小二乘法；当样本容量很小时，宜用头两序数法；当抽样次数和样本容量处于中等大小水平时，宜用最小二乘法；当抽样次数足够大时，头两序数法估计结果准确度最高，最大似然法结果偏大，最小二乘法结果偏小。

关键词: 疲劳寿命分布, 威布尔分布, 极大似然法, 最小二乘法, 头两序数法

Abstract: Fatigue life distribution is usually described with Weibull distribution. Three parameter estimation methods for fatigue life distribution are introduced in this paper. These methods are maximum likelihood method, least square method and two-ordered-statistic method. They are applied to parameter estimation of sample sets of different sample numbers and different sample sizes. The estimation results of different methods are compared. It is shown that: the maximum likelihood method and least square method are preferable when the sample number is very small; the two-ordered-statistic method is better when the sample size is quite small; the least square method is preferred when sample number and sample size are average; when the sample number is big enough, two-ordered-statistic method has the highest accuracy, maximum likelihood method leads to slightly larger results and least square method leads to moderately smaller results.