基于Bayes估计的民航维修人为错评估技术研究

摘要

This paper focuses on the human error evaluation method in civil aviation maintenance based on the Bayesian estimation theory. First of all, taking the maintenance human error probability as a random variables, meanwhile using the airline historical data as a priori information, using the 5% sites provided by the THERP Handbook as the mean probability value of maintenance human error, to determine the prior distribution model of human error using the point estimation method. Secondly, solving estiamtion method for Bayes prior distribution parameters is presented. Finally, the median posterior Beta distribution is taken as ?' s Bayesian estimation, to construct the human error probability calculation method for civil aviation maintenance. Through the example analysis on the field data, it shows that as the increase of the number of maintenance, human error probability is closer to the maximum likelihood estimation method of the failure probability. The results of case study shows the, effectiveness , accuracy and practicability of the proposed method in this paper.%研究了基于Bayes估计的民航维修人为差错评估技术.首先,将维修人为差错概率视为随机变量,利用航空公司积累的历史信息作为先验数据,以THERP手册提供的5％分位点作为维修人为差错概率中值,运用点估计法确定维修人为差错的先验分布模型.其次,给出了Bayes先验分布参数的求解方法.最后,选取验后Beta分布的中值作为θ的贝叶斯估计,构建了民航人为维修差错概率计算方法.通过实例分析,可以看出随着维修次数的增多,维修人为差错概率就越接近于极大似然法估计的失效概率.现场数据的检验反映了该评估方法的有效性、精确性和实用性.