Resolving Contentions for Complex Control Systems with Shared Resources using Robust Model Predictive Control

点击翻译按钮获取中文摘要

With the large scale of systems and data in modern industries and smart cities, shared resources, such as cloud computing, communication network, and power grid, are widely used to reduce system cost and increase maintainability. When multiple systems need access to shared resources at the same time and the total demands exceed the total supply, a contention occurs. Contentions can cause loss of information, time delay or even life-threatening safety issues. Therefore, how to schedule systems to resolve contentions is a generic problem for the control of complex systems that are coupled or connected through shared resources. To help address this problem, this project will develop transformative control techniques to co-design scheduling and control strategies for resource constrained control systems. The project combines insightful engineering with sophisticated mathematics, towards the goal of producing practically useful approaches that have rigorous performance. Once validated and transitioned to actual applications, the strategies developed by this project will have the potential to reduce congestion and pollution from traffic and transportation systems and benefit both human society and natural environment. Through a series of strongly integrated research, educational, and outreach activities, the project will also help broaden participation of underrepresented groups in STEM research.This project addresses three main questions to advance the control theory regarding to connected systems with shared resources: how to develop a general method which can co-design the scheduling and control in a unified theoretical framework; how to compute a co-design solution efficiently online for real-time systems; and how to address real-world uncertainties in the co-design problem. To answer these questions, this project will develop a robust contention-resolving model predictive control (or MPC) method as the unified theoretical framework. Analytical timing models will be established in the context of parallel machines, flow shop, and job shop scheduling to model different types of real-time control systems with shared resources. Then based on the timing models, an event-triggered MPC will be designed to achieve fast enough computation in real-time. To compensate uncertainties, the project will also develop a robust design based on forward invariant set for contention-resolving MPC to guarantee performance under system disturbances and perturbed event timing. The proposed approaches will be validated through realistic simulation software packages and field experiments to test the designs under realistic uncertainties and variations. The overall framework will be general for various applications including higher dimensional systems that are of compelling ongoing interest in engineering.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.