基于序贯回归的小样本测试性验证试验方案  

A small sample testability verification test scheme based on sequential and regression analysis

作　　者：陈然[1,2] 连光耀 张西山[1,3] CHEN Ran1,2 , LIAN Guangyao1, ZHANG Xishan1,3(1. Ordnance Technology Institute, Ordnance Engineering College, Shijiazhuang 050000, China; 2. The 75833 Unit of People＇s Liberation Army, Guangzhou 510080, China; 3. Metering Station, China Jiuquan Satellite Launch Center, Jiuquan Gansu 735000, Chin)

机构地区：[1]军械工程学院军械技术研究所,石家庄050000 [2]中国人民解放军75833部队,广州510080 [3]中国酒泉卫星发射中心计量测试站,甘肃酒泉735000

出　　处：《航空动力学报》2018年第2期305-312,共8页Journal of Aerospace Power

基　　金：预研项目(51327030104)

摘　　要：针对现有测试性验证试验方案难以实现较小样本量下满足置信水平要求的测试性评估问题,提出了基于序贯回归的小样本测试性验证试验方案。提出并证明了成败型试验序贯样本回归可行性假设;对序贯样本进行回归分析,实现了小样本下的测试性评估;建立了评估优化方案,实现了满足置信水平要求的测试性评估。实例验证表明:该方法可同时适用于有双方风险要求和置信水平要求的场合,且完所需样本量明显减少,在案例中平均减少了22.3%,最大减少了85%。A scheme based on sequential and regression analysis was presented in order to solve the problems in testability verification test schemes which might be difficult to meet the requirement of confidence level for testability evaluation while the sample size was small.First,a hypothesis of feasibility of sequential and regression analysis for the success or failure type test was presented and proved.Then,an improvement in sequential mesh test was achieved by the utilization of regression prediction estimation.Finally,the estimation of testability index satisfying the requirement of high confidence level under the condition of small sample size could be realized.Case study shows that this scheme can meet the requirement of bilateral risk and confidence level for testability evaluation,which requires smaller sample size in testability verification test.In the test of this case,the sample size could bedecreased by 22.3% on average,and by 85% at utmost.

关 键 词：测试性验证 试验方案 序贯网图 回归预测估计 小样本 

分 类 号：V233[航空宇航科学与技术—航空宇航推进理论与工程] TP114[自动化与计算机技术—控制理论与控制工程]