Statistical Inference on Part Geometry

本研究将统计理论与工程技术相结合，以建立零件检验的置信水平。目标是将以前对圆柱度的研究扩展到所有实体几何基元。该研究将采用零件几何参数的最大似然估计，例如球体的半径，来估计符合特定模型的几何形状的概率。这些模型将纳入对标称值的预期随机和系统偏差，并将在制造过程的背景下开发。一旦测量完成，所开发的理论将允许对零件几何偏差的影响进行统计表征。这项研究将能够确定几个重要的值，其中包括确保第一类和第二类错误的特定概率所需的抽样点的数目。单独检验的零件将被用来推断生产系统中剩余的零件。单个零件与未检验零件之间的关系将由公差区域的统计理论发展而来。

This research combines statistical theory with engineering technology to develop confidence levels for part inspections. The goal is to extend previous research on cylindricity to all solid geometric primitives. The research will employ maximum likelihood estimation of part geometric parameters, such as the radius of a sphere, to estimate the probability of geometry conforming to a specific model. These models will incorporate expected random and systematic deviations from nominal values and will be developed in the context of the manufacturing process. Once the measurements are complete, the theory developed will permit the statisically characterization of the impact of part geometry deviations. The research will enable the determination of several important values, among them is the number of sample points required to assure specified probabilities on type I and II errors. Individually inspected parts will be employed to draw inferences about the remainder of the parts in the production system. The relation between individual components and uninspected parts will be developed from the statistical theory of tolerance regions.