一类利用回归关系的计量指标抽样方案

A Sampling Program About Regression of Measurable Indictors

在产品质量判定的抽样检验问题中,当目标指标需用破坏性试验才能得其值时,更为常用的是用非破坏性试验可得量值的协变指标量来预报它。但在产品抽样验收问题上,未能形成理论较为严密的方法,这是由于预报误差这个关键问题的处理尚未解决得好,即给不出抽样方案的功效计算的正确或是近似性较好的公式。本文通过建立合理的数学模型,把对目标指标的质量要求化为对协变指标量的统计要求,从而利用两者的回归关系,结合两种复杂的抽样方案,给出功效函数的计算公式和计算方法,并进行了分析。

In product qualiy judgement of sample Lests, when the objective indicators get its values required destructive tasting, it is usually use the covariant index from non - destructive testing to forecast. However there are no strictress theoretics in products＇ sample test This is because we have not been well solved the kay problem which is forecast it＇s error. We can not supply the right or well approximate fomaula of sample program＇s eficacy. In this artticle we found seemly mathermtic model in order to turn objective index＇ quality requirements into Covariant index＇s ststistic requirements, so we can use their regression and two complex sample program to gain Effectiveness function＇s calculating formula and method, besides some analysis.