Collaborative Research: Robust-by-Design Networked Dynamical Systems: Bridging the Logic/Analog Divide

未来高度互连的网络控制系统的脆弱性和弹性是安全关键型应用（如自动驾驶汽车和发电/配电网）的主要问题。必须仔细研究这些关键系统的故障事件，并重新设计系统以避免它们。目前的技术水平涉及耗时和固有的不完整的设计后模拟。该项目的目的是开发理论和计算可扩展的技术，使这样的系统可以通过设计鲁棒。我们的方法可以帮助设计具有先验安全保证的更具弹性的组件和系统。我们将开发一个框架，在该框架中，网络动态系统进行分析，通过建模的事件触发和/或随机耦合的相互作用。这种动态耦合可以由邻近、远程感测或其他成对机制触发，从而导致整个系统的状态依赖的交互模式。稳定性和鲁棒性的条件将给出的L2和方差收缩的系统，和代数性质的状态相关的相互作用图。与基于Monte Carlo的方法相比，这种条件随着系统尺寸的增加而扩展。将使用该框架开发在存在外源性、对抗性和系统不确定性的情况下具有保证性能的控制算法的合成方法。该教育计划包括顶点设计演示和网络控制系统的新研究生课程。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Fragility and resiliency of future, highly-interconnected networked control system is a major concern for safety-critical applications such as autonomous vehicles and the energy generation/distribution grid. Failure events of such critical systems must be carefully studied, and systems redesigned to avoid them. The current state of the art involves time-consuming and inherently incomplete after-design simulations. The aim of this project is to develop theoretical and computationally scalable techniques so that such systems can be robust-by-design. Our methodology can help design more resilient components and systems with a-priori safety guarantees. We will develop a framework in which networked dynamic systems are analyzed by modeling interactions as event-triggered and/or stochastic couplings. Such dynamical couplings can be triggered by proximity, remote sensing or other pairwise mechanisms resulting in a state-dependent pattern of interactions for the overall system. Stability and robustness conditions will be given in terms of L2 and variance contractivity of systems, and algebraic properties of the state-dependent interaction graph. In contrast to Monte Carlo based methods, such conditions scale well with increasing system size. Synthesis methods for control algorithms with guaranteed performance in the presence of exogenous, adversarial, and system uncertainty will be developed using this framework. The education plan includes capstone design demonstrations and a new graduate level course in networked control systems.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.