Many systems operate in uncertain, possibly adversarial environments, and their successful operation is contingent upon satisfying specific requirements, optimal performance, and ability to recover from unexpected situations. Examples are prevalent in many engineering disciplines such as transportation, robotics, energy, and biological systems. This thesis studies designing correct, resilient, and optimal controllers for discrete-time complex systems from elaborate, possibly vague, specifications. The first part of the contributions of this thesis is a framework for optimal control of non-deterministic hybrid systems from specifications described by signal temporal logic (STL), which can express a broad spectrum of interesting properties. The method is optimization-based and has several advantages over the existing techniques. When satisfying the specification is impossible, the degree of violation - characterized by STL quantitative semantics - is minimized. The computational limitations are discussed. The focus of second part is on specific types of systems and specifications for which controllers are synthesized efficiently. A class of monotone systems is introduced for which formal synthesis is scalable and almost complete. It is shown that hybrid macroscopic traffic models fall into this class. Novel techniques in modular verification and synthesis are employed for distributed optimal control, and their usefulness is shown for large-scale traffic management. Apart from monotone systems, a method is introduced for robust constrained control of networked linear systems with communication constraints. Case studies on longitudinal control of vehicular platoons are presented. The third part is about learning-based control with formal guarantees. Two approaches are studied. First, a formal perspective on adaptive control is provided in which the model is represented by a parametric transition system, and the specification is captured by an automaton. A correct-by-construction framework is developed such that the controller infers the actual parameters and plans accordingly for all possible future transitions and inferences. The second approach is based on hybrid model identification using input-output data. By assuming some limited knowledge of the range of system behaviors, theoretical performance guarantees are provided on implementing the controller designed for the identified model on the original unknown system.

Power networks are one of the most complex systems in the field of electrical and computer engineering. In power networks, power supply-demand balancing can be achieved in a static or a dynamic model. In a static model, the power network cannot be easily adapted to intentional or unintentional network topology changes because the network design is predetermined, whereas in a dynamic model, the power network can be dynamically constructed and reconfigured at run-time, which leads to a more nimble, flexible, and stable system. In this dissertation, a novel Market-based Multi-agent System (MMS) is proposed to solve supply-demand balancing and power restoration problems in a dynamic model. The power network is modeled as a market environment consisting of Belief-Desire-Intention (BDI) agents representing three characters: 1) consumer, 2) supplier, and 3) middleman. The BDI agents are able to negotiate power supply and demand of the power network, with consumers exploring the market and exchanging power information with neighboring middlemen and suppliers. So long as all consumers and suppliers establish supply-demand relationships represented in tree data structures, a qualified minimal access structure is found as the lower bound of the system reliability. When contingencies occur, the agents can quickly respond and restore loads guided by the relationships using minimum computational resource. Based on case studies and simulation results, the proposed approach delivers more effective performance of contingencies response and better computation time efficiency as the scale of the power network expands. The proposed MMS shows promises for solving various real-world power supply-demand and restoration problems, and serves as a solid foundation for future power networks refinement and improvement. / Electrical and Computer Engineering

A cyber-physical control system (CPS) typically consists of a set of physical subsystems, their remote terminal units, a central control center (if applicable), and local communication networks that interconnect all the components to achieve a common goal. Applications include energy systems, autonomous vehicles, and collaborative robots. Ensuring stability, performance, and resilience in CPS requires thorough analysis and control design, utilizing robust algorithms to handle delays, communication failures, and potential cyber-attacks. Time delays are a challenge in CPS, particularly in teleoperation systems, where human operators remotely control robotic systems. These delays cause chattering, oscillations, and instability, making it difficult to achieve smooth and stable remote robot control. Applications like remote surgery, space exploration, and hazardous environment operations are highly susceptible to these disruptions. To address this issue, a novel passivity-shortage framework is proposed, that enables systems to maintain stability and transparency despite time-varying communication delays and environmental disturbances. CPS are prone to attacks, particularly Denial-of-Service (DoS) attacks, which disrupt the normal functioning of a network by overwhelming it with excessive internet traffic, rendering the communication channels unavailable to legitimate users. These attacks threaten the stability and functionality of CPS. To enhance resilience in multi-agent systems, novel distributed algorithms are proposed. These graph theory-based algorithms mitigate network vulnerabilities by incorporating strategically placed additional communication channels, thereby increasing tolerance to attacks in large, dynamic networks. The effectiveness of these proposed approaches is validated through simulations, experiments, and numerical examples. The passivity-shortage teleoperation strategies are tested using Phantom Omni devices and they show reduced chattering and better steady-state error convergence. A case study demonstrates how the proposed distributed algorithms effectively achieve consensus, even when some agents are disconnected from the network due to DoS attacks.

The electric power industry and society are facing the challenges and opportunities of transforming the present power grid into a smart grid. To meet these challenges, new types of control systems are connected over IT infrastructures. While this is done to meet highly set economical and environmental goals, it also introduces new sources of uncertainty in the control loops. In this thesis, we consider control design taking some of these uncertainties into account. In Part I of the thesis, some economical and environmental concerns in smart grids are taken into account, and a scheduling framework for static loads (e.g., smart appliances in residential areas) and dynamic loads (e.g., energy storage systems) in the distribution level is investigated. A robust formulation is proposed taking the user behavior uncertainty into account, so that the optimal scheduling cost is less sensitive to unpredictable changes in user preferences. In addition, a novel distributed algorithm for the studied scheduling framework is proposed, which aims at minimizing the aggregated electricity cost of a network of apartments sharing an energy storage system. We point out that the proposed scheduling framework is applicable to various uncertainty sources, storage technologies, and programmable electrical loads. In Part II of the thesis, we study smart grid uncertainty resulting from possible security threats. Smart grids are one of the most complex cyber-physical systems considered, and are vulnerable to various cyber and physical attacks. The attack scenarios consider cyber adversaries that may corrupt a few measurements and reference signals, which may degrade the system’s reliability and even destabilize the voltage magnitudes. In addition, a practical attack-resilient framework for networked control systems is proposed. This framework includes security information analytics to detect attacks and a resiliency policy to improve the performance of the system running under the attack. Stability and optimal performance of the networked control system under attack and by applying the proposed framework, is proved here. The framework has been applied to an energy management system and its efficiency is demonstrated on a critical attack scenario. / <p>QC 20160830</p>