Access Type

Open Access Dissertation

Date of Award

January 2025

Degree Type

Dissertation

Degree Name

Ph.D.

Department

Chemical Engineering and Materials Science

First Advisor

Helen Durand

Abstract

The fundamental objective of process control is to establish safe process operating conditions while ensuring minimal economic loss. One of the fundamental challenges of achieving this objective is non-determinism from various sources that can create unexpected process behavior. A traditional goal in industrial process control is the rejection of disturbances, which are time-varying process inputs that are not directly modified by the control system. Other sources of uncertainty can include measurement inaccuracies/noise and changes in process dynamics as a plant operates. Many techniques have been developed to ensure notions of stabilization and safety despite these sources of stochasticity for both linear and nonlinear dynamic systems. In recent decades, manufacturers have sought to push toward more efficient and digitalized operations within a framework called “smart manufacturing” or “Industry 4.0”. This framework involves an increased utilization of cyber-physical systems, which are process systems integrated with computers, communication, and networking components to improve process transparency and efficiency. Additionally, it can be accompanied by new sources of uncertainty that require a thorough characterization of safety risks and mitigation techniques. One such source of non-determinism, introduced using automation and networked systems, is a cyberattack on control systems in which a malicious actor can manipulate sensor readings or actuator outputs and potentially mask their actions. Furthermore, control being implemented on computing platforms is influenced by technological developments in the computing domain and must also be vetted from a cyber-physical systems safety perspective. For example, advances in quantum computation prompt whether this emerging computing paradigm (currently affected by hardware-related quantum “noise”) could ever be evaluated for control applications, or if the hardware limitation that introduces non-determinism as noise will render it unsafe for control applications.

Motivated by the need for a rigorous characterization of safety in manufacturing systems with new sources of non-determinism derived from digitalization and emerging technologies, our work has developed fundamental advances in control systems theory and practice toward evaluating closed-loop stability under control system cyberattacks and quantum noise in control action computation. This dissertation presents our work in these areas. First, we will present cyberattack detection schemes for nonlinear systems that provide guarantees on the time available before safety risks are potentially significant after an undetected attack on the control systems. Safety guarantees are made possible by integrating the detection policy with an optimization-based control strategy for the process called Lyapunov-based economic model predictive control (LEMPC). Second, we will utilize simulation studies to evaluate and analyze the safety associated with a linear single-input/single-output system operated by a control system that incorporates quantum computation with depolarizing error (a type of quantum noise). We will discuss the role of control system parameters and system characteristics in mitigating the effects of noise on the steady-state tracking objective.

Download

Share

COinS