基于序贯回归的小样本测试性验证试验方案

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 small sample testability verification test scheme based onsequential and regression analysis

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 be decreased by 223% on average, and by 85% at utmost.