基于虚拟样本与模糊阈值的可靠性评估方法  

Reliability Evaluation Method Based on Virtual Samples and Fuzzy Thresholds

作　　者：赵新超 吕卫民 ZHAO Xin-chao;LYU Wei-min(Naval Aviation University,Yantai 264001,China)

机构地区：[1]海军航空大学,山东烟台264001

出　　处：《火力与指挥控制》2020年第2期91-96,共6页Fire Control & Command Control

摘　　要：针对小样本、多性能参数产品在可靠性评估中存在的困难,提出一种基于虚拟样本与模糊阈值的可靠性评估方法。以性能参数距离为退化特征量,对原始样本进行虚拟增广,并利用改进Elman神经网络对虚拟样本进行修正处理;运用基于模糊阈值的可靠性评估方法对产品的可靠性进行评价;最后,以航天继电器为例对该方法的可行性进行检验,求得其在125℃条件下的寿命为137周。结果表明,该方法能够对小样本、多性能参数产品的可靠性进行更有效的评估,具有一定参考意义。In view of the difficulties in the reliability evaluation of small sample and multi-performance parameter products,a reliability evaluation method based on virtual samples and fuzzy threshold is proposed. Firstly,the original sample is virtually augmented with the performance parameter distance as the degraded feature quantity,and the virtual sample is modified by the improved Elman neural network. Then,the reliability evaluation method based on fuzzy threshold is used to evaluate the reliability of the products. Finally,the feasibility of the method is tested by aerospace relay as an example,and its life of 137 weeks at 125℃ is obtained. The result shows that this method can make a more effective evaluation of the reliability of small sample and multi-performance parameter products and has certain reference significance.

关 键 词：虚拟样本 模糊失效阈值 多性能参数 距离分析 可靠性 

分 类 号：TB114.3[理学—概率论与数理统计] TJ06[理学—数学]