Dissipativity Analysis and Resilient Design for Cyber Physical Systems

This dissertation studies dissipativity analysis and resilient design for Cyber Physical Systems (CPS) when physical entities are interconnected and controlled over communication networks. CPS incorporate heterogeneous systems in dynamic environment and provide distributed coordination intelligently. The goal of this work is to develop a compositional control infrastructure that can govern the sub-units' interactions and provide expected control performances for users. In this work, the central concept in the design of CPS is the energy based concept of passivity. Using passivity and its generalization of dissipativity, we are able to address compositionality and resilience in CPS, where resilience refers to the capability for a system to tolerances errors and takes actions when facing model uncertainties, physical disturbances, random faults, communication interference or malicious attacks.

This dissertation focus on the following problems: (1) how to quantitatively estimate and analyze the passivity of a plant from its approximate system representation considering model discrepancies; (2) how to design a joint disturbance observer and robust controller facing exogenous uncertainties and adversarial attacks; (3) how to preserve the passive and stable performances of interconnected systems over communication networks; (4) how to analyze dissipativity under a digital control framework for two-dimensional systems. This dissertation addresses complex but common problems in networked Cyber Physical Systems, the solutions of which can benefit practical applications by designing more resilient, efficient and intelligent systems.