对数稳定分布与对数正态分布相比具有厚尾等良好的特性,由于没有显式的密度函数,其参数估计有一定的困难.运用Gibbs抽样方法给出了对数稳定分布的Bayes参数估计,与对数正态分布参数估计进行了比较,并应用到恒加寿命试验数据的分析中.模拟研究说明了该方法的有效性,对数稳定分布作为新的寿命分布是合适的.
Log-stable distribution has better properties than log-normal distribution.It is difficult to estimate the parameters because of the lack of closed formulae for its density and cumulative distribution.Bayesian approach via Gibbs sampling is used to estimate the parameters in the log-stable distribution and compared with log-normal distribution.The approach is further applied to the constant stress accelerated life test.Simulation shows that the approach is efficient and the log-stable distribution is suitable as a new life time distribution.