基于多源验前信息的故障检测率Bayes融合评估方法

摘要

As the fault of the roclcet-assisted torpedo is rare in the testability and the man-made fault injection would not really estimate fault deletion rate. The prior-weighted and posterior-weighted distrubition based on the credibility of the prior information of multiple sources are set up. The fault detection rate level is estimated by bayesian fusion. The effectiveness and feasibility of the proposed method are demonstrated with a simulation example.%针对火箭助飞鱼雷测试性考核中产品自然故障很少见,注入故障的人为做法不能完全正确反映出被试品的故障检测率.采用试验多个阶段的信息,利用Bayes方法,基于可信度构建了多源验前信息的加权验前分布和验后分布,融合评估出故障检测率指标值.仿真算例表明该方法的有效性和可行性.