SGER: Identification of General Probability Models for a Taxonomy of Manufacturing Processes

这一小型探索性研究基金(SGER)的目标是：1)为制造过程分类的子集的关键质量特征创建通用概率模型。选定的分类法将规定以下分类：(A)形状和位置的几何尺寸和公差(GD&A；T)要求的类型。(B)制造工艺和产品材料的类型；及。(C)检查装置的类型。将根据模型在表示过程输出的实际行为方面的准确性和一般性来选择模型；以及2)对于每个一般概率模型，定义一个适当的汇总统计数据，该统计数据包含工程要求和过程行为的随机性质。在这项研究的第一部分，我们研究了各种材料、工艺测量设备、复杂的GD&amp；T接受区和正态/非正态单变量/多变量分布的输出，我们计划定义一组统计数据，用于总结业绩和设定目标。汇总统计数据将根据其分布特性进行选择，并将用于确定样本大小要求和关键接受值，这些值有助于支持供应商关系和设定过程性能目标。这项研究的结果将构成一套完整的方法，用于捕获和总结检查数据的行为(即，概率模型)。行业合作伙伴(英特尔)提供的数据集摩托罗拉。和Simula安全系统)将提供建模环境和研究验证。该项目的交付成果将是一份已出版的概率模型目录，这些模型可以直接映射到制造分类中条目子集的形式和位置GD&and T要求、制造工艺、材料和测量系统。如果这项研究被证明是成功的。将为分类法中的更多类别准备后续项目。总体而言，本研究项目有望为行业决策、规划和分析提供有用的模型，并在绩效汇总统计方面做出理论贡献。

The objectives of this Small Grant for Exploratory Research (SGER) are to: 1) Create general probability models for key quality characteristics of a subset of the taxonomy of manufacturing processes. The selected taxonomy will specify the following classifications (a) the type of Geometric dimensioning and tolerancing (GD&T) requirements for form and location. (b) the type of manufacturing process and product material; and (c) the type of inspection device. The models will be chosen based on their accuracy and generality in representing the realistic behavior of process outputs; and 2) For each of the general probability models, define an appropriate summary statistic that incorporates the engineering requirements and the random nature of the process behavior. Having studied the output from various materials, processes measurement devices, complex GD&T acceptance zones, and normal/non-normal univariate/multivariate distributions in the first part of this research, we plan to define a set of statistics useful for summarizing performance and setting goals. Summary statistics will be chosen based on their distributional properties and will be used to define sample size requirements and critical acceptance values useful for supporting vendor relations and setting process performance goals. The results of this research will constitute a complete methodology for capturing and summarizing the behavior of inspection data (i.e.. probability' models). Data sets provided by industrial partners (Intel. Motorola. and Simula Safety Systems) will provide the modeling environment and validation of research. The project's deliverable will be a published catalogue of probability models that can be directly mapped to form and location GD&T requirements, manufacturing processes, materials and measurement systems for a subset of entries in the manufacturing taxonomy. Should this research prove successful. a follow-on project will be prepared for more categories in the taxonomy. In total, this research project is expected to have usable models for industrial decision-making, planning, and analysis as well as the theoretical contributions in performance summary statistics.