摘要
应用极大似然法拟合样本尺寸由10到100的随机寿命数据(总样本为100)以取得二参数Weibull分布的形状参数κ,并与直接由寿命数据得到的斜度和过盈峭度相应的形状参数κ比较。同时也将寿命数据由小到大排列做同样计算。发现无序随机寿命组的极大似然法结果在样本为10到100时都给出了合理的形状参数κ及尺寸参数λ。但是,斜度和过盈峭度仅在样本大于48的场合得到相近的形状参数κ。然而,按序排列的数据只在全样本情况下得到3种方法相近的形状参数κ。随着样本由100逐渐减小到10,形状参数κ不断增大而尺寸参数λ却逐渐减小。因此认为,极大似然法、斜度和过盈峭度的形状参数κ相互重叠的出现,应是满足Weibull寿命随机特性的一个条件。
The shape parameters are evaluated for various sample size by Maximum Likelihood fit and by skewness γ_1 and excess kurtosis P2 in the life dataset （sample size = 100） randomly generated using Weibull distribution of two parameters. The same dataset is then ordered in ascent as well. Either randomly and ordered dataset are studied by gradually increasing sample size from 10 to 100. It is found that the random dataset gives shape parameter K and dimension parameter A reasonably well by Maximum Likelihood fit, but skewness and excess kurtosis give similar number only if sample size is greater than 48. However, the ordered dataset has right shape parameter K at N = 100 for all approaches, but when sample size is decreasing from 100 to 10, the shape parameter K is increasing but the dimension parameter λ is decreasing. It is found that superimposed shape parameters K by Maximum Likelihood fit and by skewness and excess kurtosis could be one of conditions of approved randomly Weibull distribution.
出处
《轴承》
北大核心
2012年第3期38-42,共5页
Bearing
