EAGER: A Self-Healing Approach for Smart Assembly Systems

摘要本基金资助制造系统品质改善新范例的初步研究。这种模式基于智能自愈系统，该系统通过在装配过程中自动纠正故障或应用补偿行动来提高生产质量。智能自愈系统将通过使用人工智能复制生物有机体的关键特征，如意识、适应、冗余和去中心化。通过开发一个可控装配试验台，探索生物机制在控制系统设计中的适用性。拟议的测试平台将集成最先进的传感器-执行器网络，以复制兼容或柔性部件的组装。提出的算法将集成三种数据源：系统预测模型、过程数据和人类知识。该项目重点研究汽车、重型设备、家电和航空航天工业中复杂装配系统的一种新的质量方法。自愈机制将分层次地在三个不同的层次上攻击故障：组件、站点和系统。一种范式转换的自我修复方法的发展将提供更快和更精确的故障预防、检测、识别和纠正。减少材料浪费和不良零件的数量，提高组装产品的质量，并通过减少制造停机时间来提高效率和生产能力。本研究的目标是在三个方面推进现有的知识：1)复杂非线性系统的预测装配变化模型；2)基于信息融合的装配系统监测-检测-诊断方法；3)采用生物启发方法的自修复装配系统的数学框架。

ABSTRACTThis grant provides funding to conduct preliminary research in a new paradigm in quality improvement for manufacturing systems. This paradigm is based on a smart self-healing system that increases production quality by autonomously correcting faults or applying compensation actions during the assembly process. The smart self-healing systems will replicate key characteristics of bio-organisms such as awareness, adaptation, redundancy, and decentralization by using artificial intelligence. The applicability of biological mechanisms in designing a control system will be explored by developing a controllable assembly testbed. The proposed testbed will integrate state-of-the-art sensor-actuator networks to replicate the assembly of compliant or flexible parts. The proposed algorithms will integrate three sources of data: system predictive models, process data, and human knowledge. This project focuses on a new quality method for complex assembly systems found in the automotive, heavy equipment, appliance, and aerospace industries. The self-healing mechanisms will hierarchically attack faults at three different levels: component, station, and system.The development of a paradigm-shifting self-healing approach to quality improvement will provide faster and more precise fault prevention, detection, identification, and correction. Material waste and the number of defective parts will be reduced, the quality of assembled products will be improved, and efficiency and production capacity will be increased through diminished manufacturing downtime. The work aims to advance the current knowledge in three areas: 1) predictive assembly variation models for complex nonlinear systems; 2) monitoring-detection-diagnosis methods for information fusion-based assembly systems; and 3) mathematical framework for a self-healing assembly system using a bio-inspired approach.