一种用于小样本合格率动态估计的Bayes方法

A Bayesian Method for Pass-Rate Estimation Based on Small Sample

针对复杂机械产品制造业广泛存在的大批量生产、小样本检测的特点 ,在采用矩法由历史测量数据构造产品合格率参数先验分布的基础上 ,提出一种综合历史数据与当前小样本信息进行合格率估计的 Bayes算法 ,建立了动态递推模型 ,实现了合格率的准确估计 .采用统计仿真的方法对动态估计精度进行了分析 ,给出了依据离线测量数据进行合格率评估的计算实例 .结果表明 ,所提出的方法是有效而且实用的 。

For the dimensional quality evaluation of complex mechanical products based on small sample, an empirical Bayesian infer algorithm for pass rate estimation was provided, which took full advantages of the historical inspection data and the information of current sample. And a dynamic recurrence estimation model for the pass rate evaluation was developed. The statistical simulation proves that the developed method is more precise than the traditional simple moving average methods. A practical example about pass rate estimation was also put forward.