Collaborative Research: Multivariate Methods for On-Line and Off-Line Quality Control and Improvement

协作研究：在线和离线质量控制和改进的多元方法

• 批准号: 9908032
• 负责人: Kwokleung Tsui
• 金额: $ 8.37万
• 依托单位: Georgia Tech Research Corporation
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1999
• 资助国家: 美国
• 起止时间: 1999-09-01 至 2002-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9908032&HistoricalAwards=false
• 关键词: Collaborative; Research; Multivariate; Methods; Line

This research is a collaborative effort with the University of Alabama-Tuscaloosa. This grant provides funding for the development and evaluation of new statistical methods to be used to monitor and improve manufacturing processes. These methods are to be applied in situations for which quality depends on several related quality measurements. The research involves both off-line and on-line methods. The off-line methods require the design of experiments and the most informative and efficient analysis of the resulting data. Emphasis is on the robust design of processes and products so that their performance is less sensitive to hard-to-control noise factors, such as environmental conditions during and after production. Robust design improves product quality, manufacturability, and reliability. The on-line methods use data collected over time to detect changes in the process for which action is required. The use of multivariate outputs results in more reliable decisions about the process. Also, the methods developed will apply in cases for which the usual assumptions, such as multivariate normality, are not valid. Attribute data, for example, will be considered. The results of this research will lead to new statistical methods for improving design of products and manufacturing processes. It will also lead to more reliable methods for monitoring the quality of processes over time to detect important process changes. This will make quality improvement activities more effective in the design, development, and production stages of manufacturing.

这项研究是与美国塔斯卡卢萨大学的合作成果。 该赠款为开发和评估用于监测和改进制造过程的新统计方法提供资金。 这些方法适用于质量取决于几个相关质量测量的情况。本研究采用了离线和在线两种方法。 离线方法需要设计实验和对所得数据进行最翔实和有效的分析。重点是工艺和产品的稳健设计，使其性能对难以控制的噪声因素不太敏感，例如生产过程中和生产后的环境条件。 稳健的设计提高了产品质量、可制造性和可靠性。在线方法使用随着时间的推移收集的数据来检测需要采取行动的过程中的变化。 多变量输出的使用导致关于过程的更可靠的决策。 此外，开发的方法将适用于通常的假设，如多元正态性，是无效的情况下。例如，将考虑属性数据。 这项研究的结果将导致新的统计方法，以改善产品设计和制造工艺。 它还将带来更可靠的方法来监控流程质量，以检测重要的流程变化。这将使质量改进活动在制造的设计、开发和生产阶段更加有效。