Ensuring reliable operation of large power systems subjected to multiple outages is a challenging task because of the combinatorial nature of the problem. Traditional methods of steady-state security assessment in power systems involve contingency analysis based on AC or DC power flows. However, power flow based contingency analysis is not fast enough to evaluate all contingencies for real-time operations. Therefore, real-time contingency analysis (RTCA) only evaluates a subset of the contingencies (called the contingency list), and hence might miss critical contingencies that lead to cascading failures.This dissertation proposes a new graph-theoretic approach, called the feasibility test (FT) algorithm, for analyzing whether a contingency will create a saturated or over-loaded cut-set in a meshed power network; a cut-set denotes a set of lines which if tripped separates the network into two disjoint islands. A novel feature of the proposed approach is that it lowers the solution time significantly making the approach viable for an exhaustive real-time evaluation of the system. Detecting saturated cut-sets in the power system is important because they represent the vulnerable bottlenecks in the network. The robustness of the FT algorithm is demonstrated on a 17,000+ bus model of the Western Interconnection (WI). Following the detection of post-contingency cut-set saturation, a two-component methodology is proposed to enhance the reliability of large power systems during a series of outages. The first component combines the proposed FT algorithm with RTCA to create an integrated corrective action (iCA), whose goal is to secure the power system against post-contingency cut-set saturation as well as critical branch overloads. The second component only employs the results of the FT to create a relaxed corrective action (rCA) that quickly secures the system against saturated cut-sets. The first component is more comprehensive than the second, but the latter is computationally more efficient. The effectiveness of the two components is evaluated based upon the number of cascade triggering contingencies alleviated, and the computation time. Analysis of different case-studies on the IEEE 118-bus and 2000-bus synthetic Texas systems indicate that the proposed two-component methodology enhances the scope and speed of power system security assessment during multiple outages.

Low frequency oscillations (LFOs) are recognized as one of the most challenging problems in electric grids as they limit power transfer capability and can result in system instability. In recent years, the deployment of phasor measurement units (PMUs) has increased the accessibility to time-synchronized wide-area measurements, which has, in turn, enabledthe effective detection and control of the oscillatory modes of the power system. This work assesses the stability improvements that can be achieved through the coordinated wide-area control of power system stabilizers (PSSs), static VAr compensators (SVCs), and supplementary damping controllers (SDCs) of high voltage DC (HVDC) lines, for damping electromechanical oscillations in a modern power system. The improved damping is achieved by designing different types of coordinated wide-area damping controllers (CWADC) that employ partial state-feedback. The first design methodology uses a linear matrix inequality (LMI)-based mixed H2/Hinfty control that is robust for multiple operating scenarios. To counteract the negative impact of communication failure or missing PMU measurements on the designed control, a scheme to identify the alternate set of feedback signals is proposed. Additionally, the impact of delays on the performance of the control design is investigated. The second approach is motivated by the increasing popularity of artificial intelligence (AI) in enhancing the performance of interconnected power systems. Two different wide-area coordinated control schemes are developed using deep neural networks (DNNs) and deep reinforcement learning (DRL), while accounting for the uncertainties present in the power system. The DNN-CWADC learns to make control decisions using supervised learning; the training dataset consisting of polytopic controllers designed with the help of LMI-based mixed H2/Hinfty optimization. The DRL-CWADC learns to adapt to the system uncertainties based on its continuous interaction with the power system environment by employing an advanced version of the state-of-the-art deep deterministic policy gradient (DDPG) algorithm referred to as bounded exploratory control-based DDPG (BEC-DDPG). The studies performed on a 29 machine, 127 bus equivalent model of theWestern Electricity Coordinating Council (WECC) system-embedded with different types of damping controls have demonstrated the effectiveness and robustness of the proposed CWADCs.

As the field of machine learning increasingly provides real value to power system operations, the availability of rich measurement datasets has become crucial for the development of new applications and technologies. This dissertation focuses on the use of time-series load data for the design of novel data-driven algorithms. Loads are one of the main factors driving the behavior of a power system and they depend on external phenomena which are not captured by traditional simulation tools. Thus, accurate models that capture the fundamental characteristics of time-series load dataare necessary. In the first part of this dissertation, an example of successful application of machine learning algorithms that leverage load data is presented. Prior work has shown that power systems energy management systems are vulnerable to false data injection attacks against state estimation. Here, a data-driven approach for the detection and localization of such attacks is proposed. The detector uses historical data to learn the normal behavior of the loads in a system and subsequently identify if any of the real-time observed measurements are being manipulated by an attacker. The second part of this work focuses on the design of generative models for time-series load data. Two separate techniques are used to learn load behaviors from real datasets and exploiting them to generate realistic synthetic data. The first approach is based on principal component analysis (PCA), which is used to extract common temporal patterns from real data. The second method leverages conditional generative adversarial networks (cGANs) and it overcomes the limitations of the PCA-based model while providing greater and more nuanced control on the generation of specific types of load profiles. Finally, these two classes of models are combined in a multi-resolution generative scheme which is capable of producing any amount of time-series load data at any sampling resolution, for lengths ranging from a few seconds to years.

The broad deployment of time-synchronized continuous point-on-wave (CPoW) modules will enable electric power utilities to gain unprecedented insight into the behavior of their power system assets, loads, and distributed renewable generation in real time. By increasing the available level of detail visible to operators, serious fault events such as wildfire-inducing arc flashes, safety-jeopardizing transformer failures, and equipment-damaging power quality decline can be mitigated in a data-driven, systematic manner. In this research project, a time-synchronized micro-scale CPoW module was designed, constructed, and characterized. This inductively powered CPoW module, which operates wirelessly by using the current flowing through a typical distribution conductor as its power source and a wireless data link for communication, has been configured to measure instantaneous line current at high frequency (nominally 3,000 samples per second) with 12-bit resolution. The design process for this module is detailed in this study, including background research, individual block design and testing, printed circuit board (PCB) design, and final characterization of the system. To validate the performance of this module, tests of power requirements, measurement accuracy, battery life, susceptibility to electromagnetic interference, and fault detection performance were performed. The results indicate that the design under investigation will satisfy the technical and physical constraints required for bulk deployment in an actual distribution network after manufacturing optimizations. After the test results were summarized, the future research and development activities needed to finalize this design for commercial deployment were identified and discussed.

Increase in the usage of Internet of Things(IoT) devices across physical systems has provided a platform for continuous data collection, real-time monitoring, and extracting useful insights. Limited computing power and constrained resources on the IoT devices has driven the physical systems to rely on external resources such as cloud computing for handling compute-intensive and data-intensive processing. Recently, physical environments have began to explore the usage of edge devices for handling complex processing. However, these environments may face many challenges suchas uncertainty of device availability, uncertainty of data relevance, and large set of geographically dispersed devices. This research proposes the design of a reliable distributed management system that focuses on the following objectives: 1. improving the success rate of task completion in uncertain environments. 2. enhancing the reliability of the applications and 3. support latency sensitive applications. Main modules of the proposed system include: 1. A novel proactive user recruitment approach to improve the success rate of the task completion. 2.Contextual data acquisition and integration of false data detection for enhancing the reliability of the applications. 3. Novel distributed management of compute resources for achieving real-time monitoring and to support highly responsive applications. User recruitment approaches select the devices for offloading computation. Proposed proactive user recruitment module selects an optimized set of devices that match the resource requirements of the application. Contextual data acquisition module banks on the contextual requirements for identifying the data sources that are more useful to the application. Proposed reliable distributed management system can be used as a framework for offloading the latency sensitive applications across the volunteer computing edge devices.

The accurate monitoring of the bulk transmission system of the electric power grid by sensors, such as Remote Terminal Units (RTUs) and Phasor Measurement Units (PMUs), is essential for maintaining the reliability of the modern power system. One of the primary objectives of power system monitoring is the identification of the snapshots of the system at regular intervals by performing state estimation using the available measurements from the sensors. The process of state estimation corresponds to the estimation of the complex voltages at all buses of the system. PMU measurements play an important role in this regard, because of the time-synchronized nature of these measurements as well as the faster rates at which they are produced. However, a model-based linear state estimator created using PMU-only data requires complete observability of the system by PMUs for its continuous functioning. The conventional model-based techniques also make certain assumptions in the modeling of the physical system, such as the constant values of the line parameters. The measurement error models in the conventional state estimators are also assumed to follow a Gaussian distribution. In this research, a data mining technique using Deep Neural Networks (DNNs) is proposed for performing a high-speed, time-synchronized state estimation of the transmission system of the power system. The proposed technique uses historical data to identify the correlation between the measurements and the system states as opposed to directly using the physical model of the system. Therefore, the highlight of the proposed technique is its ability to provide an accurate, fast, time-synchronized estimate of the system states even in the absence of complete system observability by PMUs.
The state estimator is formulated for the IEEE 118-bus system and its reliable performance is demonstrated in the presence of redundant observability, complete observability, and incomplete observability. The robustness of the state estimator is also demonstrated by performing the estimation in presence of Non-Gaussian measurement errors and varying line parameters. The consistency of the DNN state estimator is demonstrated by performing state estimation for an entire day.

Reliable operation of modern power systems is ensured by an intelligent cyber layer that monitors and controls the physical system. The data collection and transmission is achieved by the supervisory control and data acquisition (SCADA) system, and data processing is performed by the energy management system (EMS). In the recent decades, the development of phasor measurement units (PMUs) enables wide area real-time monitoring and control. However, both SCADA-based and PMU-based cyber layers are prone to cyber attacks that can impact system operation and lead to severe physical consequences.

This dissertation studies false data injection (FDI) attacks that are unobservable to bad data detectors (BDD). Prior work has shown that an attacker-defender bi-level linear program (ADBLP) can be used to determine the worst-case consequences of FDI attacks aiming to maximize the physical power flow on a target line. However, the results were only demonstrated on small systems assuming that they are operated with DC optimal power flow (OPF). This dissertation is divided into four parts to thoroughly understand the consequences of these attacks as well as develop countermeasures.

The first part focuses on evaluating the vulnerability of large-scale power systems to FDI attacks. The solution technique introduced in prior work to solve the ADBLP is intractable on large-scale systems due to the large number of binary variables. Four new computationally efficient algorithms are presented to solve this problem.

The second part studies vulnerability of N-1 reliable power systems operated by state-of-the-art EMSs commonly used in practice, specifically real-time contingency analysis (RTCA), and security-constrained economic dispatch (SCED). An ADBLP is formulated with detailed assumptions on attacker's knowledge and system operations.

The third part considers FDI attacks on PMU measurements that have strong temporal correlations due to high data rate. It is shown that predictive filters can detect suddenly injected attacks, but not gradually ramping attacks.

The last part proposes a machine learning-based attack detection framework consists of a support vector regression (SVR) load predictor that predicts loads by exploiting both spatial and temporal correlations, and a subsequent support vector machine (SVM) attack detector to determine the existence of attacks.

Energy is one of the wheels on which the modern world runs. Therefore, standards and limits have been devised to maintain the stability and reliability of the power grid. This research shows a simple methodology for increasing the amount of Inverter-based Renewable Generation (IRG), which is also known as Inverter-based Resources (IBR), for that considers the voltage and frequency limits specified by the Western Electricity Coordinating Council (WECC) Transmission Planning (TPL) criteria, and the tie line power flow limits between the area-under-study and its neighbors under contingency conditions. A WECC power flow and dynamic file is analyzed and modified in this research to demonstrate the performance of the methodology. GE's Positive Sequence Load Flow (PSLF) software is used to conduct this research and Python was used to analyze the output data.

The thesis explains in detail how the system with 11% of IRG operated before conducting any adjustments (addition of IRG) and what procedures were modified to make the system run correctly. The adjustments made to the dynamic models are also explained in depth to give a clearer picture of how each adjustment affects the system performance. A list of proposed IRG units along with their locations were provided by SRP, a power utility in Arizona, which were to be integrated into the power flow and dynamic files. In the process of finding the maximum IRG penetration threshold, three sensitivities were also considered, namely, momentary cessation due to low voltages, transmission vs. distribution connected solar generation, and stalling of induction motors. Finally, the thesis discusses how the system reacts to the aforementioned modifications, and how IRG penetration threshold gets adjusted with regards to the different sensitivities applied to the system.

In this work, the stacked values of battery energy storage systems (BESSs) of various power and energy capacities are evaluated as they provide multiple services such as peak shaving, frequency regulation, and reserve support in an ‘Arizona-based test system’ - a simplified, representative model of Salt River Project’s (SRP) system developed using the resource stack information shared by SRP. This has been achieved by developing a mixed-integer linear programming (MILP) based optimization model that captures the operation of BESS in the Arizona-based test system. The model formulation does not include any BESS cost as the objective is to estimate the net savings in total system operation cost after a BESS is deployed in the system. The optimization model has been formulated in such a way that the savings due to the provision of a single service, either peak shaving or frequency regulation or spinning reserve support, by the BESS, can be determined independently. The model also allows calculation of combined savings due to all the services rendered by the BESS.

The results of this research suggest that the savings obtained with a BESS providing multiple services are significantly higher than the same capacity BESS delivering a single service in isolation. It is also observed that the marginal contribution of BESS reduces with increasing BESS energy capacity, a result consistent with the law of diminishing returns. Further, small changes in the simulation environment, such as factoring in generator forced outage rates or projection of future solar penetration, can lead to changes as high as 10% in the calculated stacked value.

The grand transition of electric grids from conventional fossil fuel resources to intermittent bulk renewable resources and distributed energy resources (DERs) has initiated a paradigm shift in power system operation. Distributed energy resources (i.e. rooftop solar photovoltaic, battery storage, electric vehicles, and demand response), communication infrastructures, and smart measurement devices provide the opportunity for electric utility customers to play an active role in power system operation and even benefit financially from this opportunity. However, new operational challenges have been introduced due to the intrinsic characteristics of DERs such as intermittency of renewable resources, distributed nature of these resources, variety of DERs technologies and human-in-the-loop effect. Demand response (DR) is one of DERs and is highly influenced by human-in-the-loop effect. A data-driven based analysis is implemented to analyze and reveal the customers price responsiveness, and human-in-the-loop effect. The results confirm the critical impact of demographic characteristics of customers on their interaction with smart grid and their quality of service (QoS). The proposed framework is also applicable to other types of DERs. A chance-constraint based second-order-cone programming AC optimal power flow (SOCP-ACOPF) is utilized to dispatch DERs in distribution grid with knowing customers price responsiveness and energy output distribution. The simulation shows that the reliability of distribution gird can be improved by using chance-constraint.