GOALI: Next generation feature-based process monitoring for smart manufacturing

过程监控的目标是检测可能导致制造环境偏离其所需操作的起因和潜在原因。如果在早期发现并纠正潜在的故障和故障，工厂的停机时间在五年内可减少50%，十年内可减少90%。然而，实现这些目标仍然非常具有挑战性，因为当前最先进的过程监控解决方案在处理先进制造过程的过程动力学和非线性方面存在局限性。智能制造流程产生的大数据集带来了额外的挑战。在这个项目中，来自奥本大学和Praxair的研究团队将开发和验证下一代基于特征的统计过程监控(SPM)框架，以此作为应对当前过程监控挑战的有效方法。拟议的项目将系统地研究各种特征和过程特征之间的潜在联系，这将为拟议的基于特征的SPM框架奠定基础。在工业物联网(IIoT)仍处于初级阶段的情况下，研究团队渴望开发实验室规模的IIoT支持制造技术试验台(MTT)，它将允许详细了解IIoT传感器的动态行为，并建立模拟模型以准确捕获IIoT传感器的行为。该团队计划开发一套支持IIoT的模拟MTT和一个全面的特征库、用于指导相关特征识别的关联决策树，以及用于补充基于特征的SPM框架的自动特征选择算法。支持IIoT的MTT型模拟器套件和特征库将以开放源代码的形式公开提供。所提出的基于特征的监测方法可以扩展到其他领域，如基于特征的控制、基于特征的优化和基于特征的预测性维护。拟议的教育和外展工作重点是为学生在先进制造业中的职业生涯做准备，并为少数民族提供研究机会。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The goal of process monitoring is to detect the onset and identify the underlying reasons that can cause a manufacturing environment to deviate from its desired operation. If potential faults and failures are detected and corrected while still incipient, reduction in plant downtimes of up to 50% in five years and up to 90% in ten years can be achieved. However, achieving these targets remains very challenging, as the current state-of-the-art process monitoring solutions have limitations in addressing, for example, the process dynamics and nonlinear of advanced manufacturing processes. The big data sets that are generated from smart manufacturing processes pose additional challenges. In this project a research team from Auburn University and Praxair will develop and validate a next-generation feature-based statistical process monitoring (SPM) framework as an effective way to address current challenges in process monitoring.The proposed project will systematically examine the underlying connections between various features and process characteristics, which will lay the foundation for the proposed feature-based SPM framework. With the industrial Internet-of-things (IIoT) still in its infancy, the research team aspires to develop lab scale IIoT-enabled manufacturing technology testbeds (MTT), which will allow detailed understanding of the dynamic behavior of IIoT sensors, and simulation models to accurately capture the behavior of IIoT sensors. The team plans to develop a suite of simulated IIoT-enabled MTTs and a comprehensive feature library, the associated decision tree to guide the relevant feature identification, and the automated feature selection algorithm to complement the feature-based SPM framework. The suite of IIoT-enabled MTT simulators and the feature library will be made publicly available in the form of open source codes. The proposed feature-based monitoring methodology can be extended to other areas, such as feature-based control, feature-based optimization and feature-based predictive maintenance. The proposed educational and outreach efforts focus on preparing students for careers in advanced manufacturing and providing research opportunities to minorities.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.