Cyber-resilient Control Structures in DC Microgrids with Cyber-Physical Threats

In recent years, the power system has been undergoing increasing pressure to expand its capacity in line with the growth of electricity demand, which has posed significant difficulties for many power industry sectors. Although power generation technologies are improving rapidly, these challenges cannot be overcome without integrating renewable energy sources (RES) into the power grid. RESs, such as PV panels, wind turbines, etc., and energy storage technologies can be integrated into the electricity system as a local power generation solution. There has been a growing awareness of the importance of local power generation in recent years, resulting in the emergence of the microgrid (MG) concept. MGs are promising solutions for improving power system efficiency and resilience, thanks to their versatility and controllability. In various applications, DC microgrid (DCMG) distribution systems are preferred over conventional AC microgrid (ACMG) systems for several reasons. These include a simpler control system as most RESs have DC outputs, easier integration for RESs, and elimination of some of the most challenging issues in ACMGs, such as controlling the reactive power. DCMGs have many potential uses, including but not limited to distribution systems, data centers, electric ships, and transportation. Many concerns regarding the resiliency, reliability, and stability of DCMGs control have been raised in recent years due to the increased penetration of RESs into traditional power grids. To overcome these challenges, coordinated control of power generation resources, energy storage systems, and loads is necessary. Some of the crucial control tasks in DCMGs are balancing the state of charge of batteries, sharing the load current proportionally, regulating the DC bus voltage, and detecting and isolating faults, among others. Due to the diverse control tasks and their different time frames, hierarchical control schemes have received significant attention for DCMGs. In this structure, control tasks are commonly organized into three levels, namely primary, secondary, and tertiary control levels, where primary droop-based controllers are typically decentralized, while secondary and tertiary controllers are normally centralized. Distributed control schemes rely heavily on communication networks and are finding increasing use in the hierarchical control of MGs. Specifically, in DCMGs, the hierarchical control structure with distributed control strategies has been frequently implemented and attracted considerable attention compared with other control structures.

Although distributed control systems have many benefits, one major concern is their vulnerability to malicious intrusion due to heavy reliance on data transmission via communication links. Consequently, there is significant interest in developing and employing attack detection strategies for DCMGs that make use of distributed control systems; however, there are still substantial technological gaps that need to be addressed.

This thesis investigates the identification and mitigation of cyber-resiliency challenges faced by DCMG. In this regard, several innovative methods to enhance the resiliency of DCMGs against various types of manipulation, ranging from unknown disturbances to false data injection (FDI) cyber-attacks, are proposed. In the proposed methods, the type and place of intrusion of cyber-physical threats are considered. The proposed control methods focus on improving system resilience in both the cyber and physical layers. This is particularly important in the presence of disturbance injections at the primary level of the hierarchical control system. Thus, the proposed methods can be classified as follows based on their level of operation in the hierarchical control system of DCMGs:

Secondary level

· This project aims to enhance system resilience by developing a data-driven framework in the cyber layer of the secondary control system of DCMGs. The proposed framework focuses on detecting and mitigating FDI attacks. First, the intrusion is detected, then the manipulated signal is amended, and the approximation of the pre-attack value is provided to the control system to keep the system performance within a safe range. In the proposed method, the output voltage and current of the target unit are predicted using real-time machine learning (ML)--based estimators. The residual signal calculated from the real and predicted voltage and current values is then analyzed. The goal is to monitor the state of information exchange in communication links and to determine whether the system is under malicious attack. An online change point detection method is considered to detect any unusual change point in the error signals, which raises the alarm for the presence of an attack in the cyber layer. Moreover, by utilizing a mitigation method, the secondary controllers, instead of receiving a manipulated signal, will receive an amended signal.

· A distributed secondary controller for DCMGs is proposed in this project, which reduces dependence on the information from neighboring units to achieve voltage consensus, current sharing, and reference voltage tracking. Distributed control schemes for DCMG systems rely heavily on data, which can negatively impact their cyber resilience. However, our proposed distributed control method can achieve the same performance with reduced reliance on data transfer. This is achieved by using a distributed finite-time secondary controller from the literature and leveraging the physical equations in the DCMG network to eliminate the need for voltage information from neighboring units. Local measurements of load, corresponding unit currents, and line resistances are relied upon to achieve voltage consensus. The control law for the unit responsible for tracking reference voltage in an interconnected network setting is modified, freeing it from other responsibilities. Finally, a saturation function is included in the secondary controller with an integrator anti-windup logic to ensure system voltages remain at safe levels.

Primary level

· To cope with the security challenges originating from the primary layer, two different model-based voltage control schemes are proposed in this thesis for local controllers at the primary level of the hierarchical control structures. The first can tackle a wide range of unknown external disturbances and fulfill the primary level control objectives, such as tracking the desired voltage setpoints received from the secondary controller.

· An improved robust voltage control strategy for DC-DC power converters is also proposed in this thesis that can accurately track voltage setpoints, even in the presence of measurement noise, delays, model parameter uncertainties, and external disturbances. This task is challenging for DC-DC power converters in DCMGs, as the load changes occur instantaneously. By utilizing the proposed scheme, the system can achieve more reliable voltage tracking with lower tracking errors, resulting in improved system performance within the standard range set by the IEEE.