Presidential Young Investigator Award: Feedback Structures for Supervisory Control

这项研究的目的是开发设计和评估感觉反馈回路的方法，用于柔性制造单元中的监督决策和控制。在这些应用中，在连续空间和时间内演化的过程被建模和控制为异步离散事件。将待自动化的任务分解为基本的离散操作，子任务之间具有交替排序和优先关系。为了分析可供选择的反馈结构，正在探索信息论措施，以此作为表示业务需求以及系统组成部分的信息和决策空间的一种手段。基于信息的模型将表征传感器分辨率、执行器精度以及信息处理和决策算法的替代内部模型的影响。来自最优随机控制的技术正在被研究，以潜在地应用于具有实时资源分配的灵活小区的决策问题。作为评估可选反馈结构和决策算法的工具，将实现自动化过程的离散事件仿真模型，并显式表示感觉反馈回路。

The objective of this research is to develop methods for design and evaluation of sensory feedback loops for supervisory decision and control in flexible manufacturing cells. In these applications, processes evolving in continuous space and time are modeled and controlled as asynchronous discrete events. The task to be automated is decomposed into basic discrete operations with alternative sequencing and precedence relations among the subtasks. To analyze alternative feedback structures, information-theoretic measures are being explored as a means for representing the requirements of the operations and the information and decision spaces of the system components. Information-based models will characterize sensor resolution, actuator accuracy, and the effects of alternative internal models for information processing and decision algorithms. Techniques from optimal stochastic control are being investigated for potential application to decision problems in flexible cells with real-time resource allocation. As a tool for evaluating alternative feedback structures and decision algorithms, a discrete-event simulation model of the automated processes will be implemented with explicit representation of the sensory feedback loops.