Design of Experiments (DOE) Based Automatic Process Control (APC): A Methodology for Process Variation Reduction Beyond Robust Parameter Design

该项目致力于开发基于实验设计(DOE)回归模型和噪声因素实时测量或估计的自动过程控制(APC)方法。这项研究的中心思想是通过整合能源部、统计过程控制和控制与评估理论的学科，开发一种实现自动化过程控制的方法学。研究的基本问题包括：(1)适用于APC的可控因素和噪声因素的新分类；(2)考虑系统可辨识性的用于过程建模的DOE的测试设计和分析；(3)考虑DOE回归模型和噪声因素估计的不确定性的谨慎控制策略；(4)DOE模型的在线更新和带监督的自适应控制。这项研究扩展了能源部、SPC和APC三个成熟学科的现有理论，形成了一种新的基于能源部的APC方法论。这一新方法与稳健设计和SPC相结合，将为广泛的制造工艺提供更有效的技术。这类工艺的例子包括冲压、锻造、半导体和复合材料制造工艺。

This project focuses on the development of automatic process control (APC) methodologies based on Design of Experiments (DOE) regression models and real-time measurement or estimation of noise factors. The central idea of the proposed research is to develop a methodology to achieve automatic process control by integrating the disciplines of DOE, SPC, and control and estimation theory. Various fundamental issues will be studied, which include: (1) a new classification of controllable factors and noise factors applicable to serve as feedback information for APC; (2) test design and analysis of DOE for process modeling with consideration of system identifiability for the control purpose; (3) cautious control strategy with consideration of uncertainties in DOE regression models and noise factor estimation; and (4) on-line DOE model updating and adaptive control with supervision.The success of the research will lead to a new scientific basis and practical tools for designing and implementing APC in complex manufacturing processes. The research expands the existing theory in each of the three well-developed disciplines of DOE, SPC, and APC to form a new "DOE-based APC" methodology. This new methodology, in conjunction with robust design and SPC, will provide more effective techniques for a broad range of manufacturing processes. Examples of such processes include stamping, forging, semiconductor, and composite material manufacturing processes.