指数型产品多阶段可靠性增长的Bayes模型

基于D-S证据理论融合验前信息,建立了一种适用于小子样复杂系统多阶段可靠性增长分析的Bayes模型。选择Gamma分布作为失效率的先验分布,通过多源可靠性信息融合,将专家经验转换成概率分布, 利用D-S证据理论融合多个专家信息,确定了先验分布参数,结合产品研制阶段试验数据,根据Bayes统计推断理论,给出了失效率、平均故障间隔时间（MTBF）和可靠度的Bayes点估计和置信下限。以兆瓦级直驱式风力发电机研制试验验证了该模型的有效性。

Bayes Model of Exponential Product Multi-stage Reliability Growth

A multi-stage   reliability  growth  Bayes  model  was  built based on  D-S  evidence  theory and prior information,  that  was  adaptive  to  small-sample complicated  systems.Gamma distribution was used as prior distribution of failure rate.Through integrated usage of multiple-source  reliability informations,the  expertise  was converted into probability distribution.Using D-S evidence theory and multiple experts' informations,the prior distribution parameters were known.With  test  data  at  product  development  and  Bayes  statistical inference theory,the  failure rate,MTBF,point estimation and lower confidence were obtained.The  validity of this approach was illustrated by the development  and  test  of  MW  direct-drive  wind turbine.