摘要
为解决高价值弹药可靠性鉴定中试验样本量大、成本高的问题,提出了基于小子样变动统计思想的可靠性综合评定方法。考虑到弹药可靠性在研制阶段上呈现动态特性、环境上呈现差异特性和相似装备上呈现关联特性,基于小子样变动统计思想建立了弹药可靠性综合评定模型;针对多源历史样本的“异母体”问题,以信息熵和条件熵理论对无历史样本下和有历史样本下可靠度的不确定性进行量化,以不确定度减少比率作为该历史样本相较于现场样本的继承因子;基于贝塔分布函数和继承因子构建了关于可靠度的混合先验分布,由贝叶斯理论推导得到可靠度的验后分布,从而实现对高价值弹药的高效可靠性评定。以某型弹药为应用对象,验证了本文方法的有效性和合理性。
To solve the problem of large sample and high cost of high-value ammunition on reliability qualification test,the synthetic reliability assessment was proposed based on small-sample dynamic population thought.For the reliability of ammunition,considering the dynamic characteristic present on research stage,the differential characteristic present on environment and relevance characteristic on similar equipment,the synthetic reliability assessment of ammunition was established based on small-sample dynamic population thought;Regarding the different maternal body problem,the uncertainty of reliability with historical sample and without historical sample was quantified by using the information entropy and conditional entropy theory,and the inheritance factor of historical sample to current sample was expressed with uncertainty-reducing ratio.The mixed prior probability distribution of reliability was constructed based on beta distribution function and inheritance factor,and the posterior probability distribution was derived by Bayesian theory,finally achieving the efficient reliability assessment for high-value ammunition.The validity and rationality of above method was tested and verified through taking certain ammunition as application object.
作者
张雷雷
解龙
高旭
孙天宇
张峰
ZHANG Leilei;XIE Long;GAO Xu;SUN Tianyu;ZHANG Feng(Xi’an Institute of Control Technology, Xi’an 710065, China;School of Methanics, Civil Engineering and Architecture, Northwestern Polytechnical University, Xi’an 710129, China)
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2021年第11期3399-3404,共6页
Systems Engineering and Electronics
关键词
高价值弹药
可靠性综合评定
小子样变动统计
贝叶斯
信息熵
high-value ammunition
synthetic reliability assessment
small-sample dynamic population statistics
bayesian theory
information entropy
