Supervisory Control in Automated Manufacturing Processes: The Design of Human-Computer Interaction for "Lights-Out Factories"

大多数关于制造自动化的研究集中在全自动化控制和调度技术的推导上。“熄灯”自动化的另一种更现实的范例是监督控制；制造控制和调度流程的设计，明确地将人工决策者与底层自动化集成在一起。监控设计人机交互，以增强和扩展人的作用和决策的有效性。监督控制的目标和意想不到的副作用都不是将人类决策者从系统中自动化，也不是将人类的角色减少到一组不受欢迎或无效的任务。同时，监督控制改善了自动化控制固有的灵活性局限性。本研究提供了监督控制作为一个替代的目标，在制造过程中的完全自动化。通过监督控制，目标是在控制过程中设计利用人的技能并提高人的效率和整体系统性能的人的功能。本研究提出了三部分方法来探索先进制造系统中的监控：首先，开发一个交互式的、实时的、面向对象的制造模拟器；其次，开发建模方法来规定/描述控制和排除制造过程故障的操作员；第三，利用操作者模型和面向对象的制造模拟器系统探索，结合现实的制造系统动力学，将分析方法和知识技术相结合，有效地设计出在线决策支持系统。

Most research on manufacturing automation focuses on the derivation of fully automated control and scheduling techniques. An alternative and more realistic paradigm to "lights out" automation is supervisory control; the design of manufacturing control and scheduling processes that explicitly integrate human decision makers with underlying automation. Supervisory control designs the human-computer interaction in order to augment and to extend the human's role and decision making effectiveness. Neither the goal nor the unintended side-effects of supervisory control are to automate the human decision maker out of the system nor to reduce the human's role to a set of undesirable or ineffective tasks. At the same time, supervisory control ameliorates the inherent limitations of flexibility in automated control. This research offers supervisory control as an alternative to the goal of full automation in manufacturing processes. With supervisory control, the goal is to design into the control process human functions that utilize human skills and that enhance human effectiveness and overall system performance. The research offers a three part approach to explore supervisory control in advanced manufacturing systems: first, the development of an interactive, real-time, object-oriented manufacturing simulator; second, the development of modeling methods to prescribe/describe operators who control and troubleshoot manufacturing processes; and third, to use the operator models and the object-oriented manufacturing simulator to systematically explore, with realistic manufacturing system dynamics, the integration of analytical methods and knowledge-based techniques, effective designs for an on-line decision support system.