Systems Design and Analysis: Stability, Performance and Robustness

9403742 Meyn这项建议涉及开发和应用分析复杂系统的新方法，其中大多数可视为具有反馈的一般多类网络系统的特例。也许最重要的例子是半导体制造系统的设计，其中分析的一个影响将包括构建稳定和高效的路线和调度策略；但理解一般系统的应用要广泛得多。具体地说，我们建议对系统分析的三个方面进行研究：(I)评估系统稳定性和不稳定性的方法，其中不仅包括系统的离散和连续时间表示，还包括此类系统的流体和扩散近似；(Ii)稳定系统的性能度量，特别是此类系统达到稳定状态的速率比较；(Iii)系统设计对假设扰动的稳健性。所设想的具体类型的应用包括制造系统和更一般的带反馈的排队系统，所要开发的稳定性和性能结果非常适合这些系统。我们还将详细考虑这类系统的调度算法，并开发用于建模和性能验证的仿真方法。总体而言，预计该项目将提供新的方法，将在所有(I)-(3)中统一目前的工作；并为制造和工程系统提供一些重要的应用程序，这两个系统本身将是改进设计准则的重要因素。***

9403742 Meyn This proposal concerns development and application of new methodologies for analysis of complex systems, most of which can be seen as special cases of general multiclass network systems with feedback. Perhaps the most important examples is the design of semiconductor manufacturing systems, where one impact of the analysis would include the construction of stable and efficient routing and scheduling policies; but the applications of understanding the general systems is far more widespread. Specifically we propose to carry out research into three aspects of system analysis: (i) Methods of evaluating system stability and instability, which will include not only discrete and continuous time representations of systems, but also fluid and diffusion approximations for such systems; (ii) Performance measures on stable systems, and specifically comparison of rates at which such systems achieve a stable regime; (iii) Robustness of system design against perturbations in assumptions. The specific types of applications envisaged include manufacturing systems, and more general queuing systems with feedback, for which the stability and performance results to be developed are well suited. We will also consider in detail scheduling algorithms for such systems, and develop simulation methods for modeling and performance validation. Overall this project can be expected to deliver new methodologies which will unify the current work in all of (i)- (iii); and to provide a number of significant applications to manufacturing and engineering systems which, of themselves, will be a major importance in improving design guidelines. ***