Nondeterministic Control of Discrete-Event Systems for Language or (Bi) Simulation Equivalence Specifications

本研究的目的是针对基于语言/模拟/互模拟等价规范的离散事件系统的不确定性控制进行全面的研究。以往的监督控制工作主要考虑确定性监督者。对于任何最终用户来说，监督者是确定性的还是不确定性的并不重要，重要的是它是否完成了控制任务。事实上，对于不确定的对象(不确定性是由未建模的动态引起的，可能是有意的，也可能是无意的)，基于仿真/互模拟等价的规范更自然，而不确定控制可能是执行它们的唯一选择。此外，即使对于基于语言等价性的规范，控制的不确定性也可以提供几个优势(正如我们最近的工作所报告的那样)。这项建议旨在加深对不确定性控制为语言/模拟/互模拟等值的监督控制所具有的优势的理解。该项目包括位于我们控制实验室的两个教育离散事件“工厂”的实验。一个是制造系统，另一个是化学加工厂。我们还建议扩展我们现有的监控软件，以包括不确定控制综合的新算法。该项目更广泛的影响在于其离散事件系统的应用领域，包括制造、电信和网络、交通和运输、化工过程控制、嵌入式系统控制、异步硬件验证和软件验证。随着事件驱动系统的广泛适用性，培养具有离散事件背景的研究生是很重要的，这项提议将涉及2名博士生。PI每年教授一门关于离散事件系统的课程，通过该课程将传播新的结果。

The objective of this research is a comprehensive study of nondeterministic control of discrete event systems for language/simulation/bisimulation equivalence based specifications. Past work on supervisory control considered mainly the deterministic supervisors. To any end-user it does not matter whether the supervisor is deterministic or nondeterministic rather whether it fulfills the control task. In fact for nondeterministic plants (nondeterminism arises due to unmodeled dynamics, which may be intentional or unintentional), simulation/bisimulation equivalence based specification are more natural, and nondeterministic control may be the only option to enforce them. Also, even for language equivalence based specifications nondeterminism of control can offer several advantages (as reported in our recent work). The proposal is aimed toward developing further understanding of the advantages the nondeterministic control has to offer for supervisory control for language/simulation/bisimulation equivalence. The project includes experimentation with two educational discrete-event "plants" located in our controls laboratory. One is a manufacturing system whereas other is a chemical-process plant. We also propose to extend our existing supervisory control software to include the new algorithms of nodeterministic control synthesis. The broader impacts of the project lie in it's application areas of discrete-event systems including, manufacturing, telecommunications and networking, traffic and transportation, chemical process control, embedded system control, asynchronous hardware verification, and software verification. With such a wide spread applicability of event-driven systems, it is important to educate graduate students with discrete-event background, and this proposal will involve 2 PhD students. The PI teaches a course on Discrete-Event Systems each year through which the new results will be disseminated.