In recent years, the adoption of Distributed Energy Resources (DERs) in power systems has been increasing, driven by technological advancements, development of monitoring and control techniques, policy guidance among various countries, and the benefits DERs bring to the power system. These benefits include low-cost energy production, environmental sustainability promotion, and enhanced operational efficiency of the power system. For instance, demand response (DR) can alleviate pressure during peak load periods, while solar PV units and wind turbines with smart inverters can improve grid reliability through grid regulation based on IEEE Standard 1547. Despite the opportunities DERs present, their adoption also poses challenges. The growing reliance on renewable sources introduces uncertainty, variability, and intermittency, directly impacting system stability and efficiency. Addressing these challenges necessitates comprehensive research to enhance stability, improve system operations, and maximize resource utilization. This dissertation concentrates on two primary research areas: analyzing prosumer (consumers and producers, as one) consumption behavior and developing AC optimal power flow (ACOPF) models. Firstly, understanding prosumer consumption behavior is important for reducing DERs' uncertainty, particularly DR programs. This study employs a proposed probabilistic algorithm to analyze the causal relationships between prosumer consumption behavior and other factors. Two causal-oriented approaches are utilized to establish accurate prediction models and assess demand flexibility. Causal artificial intelligence facilitates intervention and counterfactual analyses of prosumers’ DR participation and consumption behavior. Finally, a Conditional Hidden Semi-Markov Model (CHSMM) is applied to model and predict household appliance electricity consumption, further enhancing understanding of prosumer behavior. Secondly, the dissertation investigates optimization models for efficient, cost-effective power system operation and resource utilization maximization. A convex two-stage socially-aware and risk-aware Second-Order Cone Programming (SOCP)-based ACOPF model is introduced to mitigate DER uncertainty, enhance PV energy utilization, and reduce operational costs. Additionally, a convex SOCP-based ACOPF model is presented for three-phase unbalanced distribution systems, incorporating the Q-V characteristics of PV units with smart inverters based on IEEE Standard 1547. This model enables the participation of PV units with smart inverters in grid voltage regulation, enhancing power system stability and achieving efficient, cost-effective operation.

The integration of Distributed Energy Resources (DER), including wind energy and photovoltaic (PV) panels, into power systems, increases the potential for events that could lead to outages and cascading failures. This risk is heightened by the limited dynamic information in energy grid datasets, primarily due to sparse Phasor Measurement Units (PMUs) placement. This data quality issue underscores the need for effective methodologies to manage these challenges. One significant challenge is the data gaps in low-resolution (LR) data from RTU and smart meters, hindering robust machine learning (ML) applications. To address this, a systematic approach involves preparing data effectively and designing efficient event detection methods, utilizing both intrinsic physics and extrinsic correlations from power systems. The process begins by interpolating LR data using high-resolution (HR) data, aiming to create virtual PMUs for improved grid management. Current interpolation methods often overlook extrinsic spatial-temporal correlations and intrinsic governing equations like Ordinary Differential Equations (ODEs) or Differential Algebraic Equations (DAEs). Physics-Informed Neural Networks (PINNs) are used for this purpose, though they face challenges with limited LR samples. The solution involves exploring the embedding space governed by ODEs/DAEs, generating extrinsic correlations for initial LR data imputation, and enforcing intrinsic physical constraints for refinement. After data preparation, event data dimensions such as spatial, temporal, and measurement categories are recovered in a tensor. To prevent overfitting, common in traditional ML methods, tensor decomposition is used. This technique merges intrinsic and physical information across dimensions, yielding informative and compact feature vectors for efficient feature extraction and learning in event detection. Lastly, in grids with insufficient data, knowledge transfer from grids with similar event patterns is a viable solution. This involves optimizing projected and transferred vectors from tensor decomposition to maximize common knowledge utilization across grids. This strategy identifies common features, enhancing the robustness and efficiency of ML event detection models, even in scenarios with limited event data.

Fault detection is an integral part for power systems as without its proper study, analysis and mitigation, people will not be able to power the various appliances and equipment required in all aspects of life. One such type of fault which is very criticalin an electrical cable but very difficult to spot is incipient fault. These momentary faults are observed for very short periods however, if it persists, this would lead to consequences like insulation wear-off and even, power outages. With the advent of machine learning in the power systems fraternity, this paper also uses a new and updated Active Learning algorithm to detect incipient fault data from a simulated test case. The objective of the paper is to detect the fault data accurately using this new and precise method. For purposes of data collection and training of the model, MATLAB Simulink and Python programming has been used respectively.

The increase in the photovoltaic (PV) generation on distribution grids may cause reverse power flows and challenges such as service voltage violations and transformer overloading. To resolve these issues, utilities need situational awareness, e.g., PV-feeder mapping to identify the potential back-feeding feeders and meter-transformer mapping for transformer overloading. As circuit schematics are outdated, this work relies on data. In cases where the advanced metering infrastructure (AMI) data is unavailable, e.g., analog meters or bandwidth limitation, the dissertation proposes to use feeder measurements from utilities and solar panel measurements from solar companies to identify PV-feeder mapping. Several sequentially improved methods based on quantitative association rule mining (QARM) are proposed, where a lower bound for performance guarantee is also provided. However, binning data in QARM leads to information loss. So, bands are designed to replace bins for increased robustness. For cases where AMI data is available but solar PV data is unavailable, the AMI voltage data and location data are used for situational awareness, i.e., meter-transformer mapping, to resolve voltage violation and transformer overloading. A density-based clustering method is proposed that leverages AMI voltage data and geographical information to efficiently segment utility meters such that the segments comprise meters of few transformers only. Although it is helpful for utilities, it may not directly recover the meter-transformer connectivity, which requires transformer-wise segmentation. The proposed density-based method and other past methods ignore two common scenarios, e.g., having large distance between a meter and parent transformer or high similarity of a meter's consumption pattern to a non-parent transformer's meters. However, going from meter-meter can lead to the parent transformer group meters due to the usual observation that the similarity of intra-cluster meter voltages is usually stronger than the similarity of inter-cluster meter voltages. Therefore, performance guarantee is provided via spectral embedding with voltage data under reasonable assumption. Moreover, the assumption is partially relaxed using location data. It will benefit the utility in many ways, e.g., mitigating voltage violations by transformer tap settings and identifying overloaded transformers.

Modern physical systems are experiencing tremendous evolutions with growing size, more and more complex structures, and the incorporation of new devices. This calls for better planning, monitoring, and control. However, achieving these goals is challenging since the system knowledge (e.g., system structures and edge parameters) may be unavailable for a normal system, let alone some dynamic changes like maintenance, reconfigurations, and events, etc. Therefore, extracting system knowledge becomes a central topic. Luckily, advanced metering techniques bring numerous data, leading to the emergence of Machine Learning (ML) methods with efficient learning and fast inference. This work tries to propose a systematic framework of ML-based methods to learn system knowledge under three what-if scenarios: (i) What if the system is normally operated? (ii) What if the system suffers dynamic interventions? (iii) What if the system is new with limited data? For each case, this thesis proposes principled solutions with extensive experiments. Chapter 2 tackles scenario (i) and the golden rule is to learn an ML model that maintains physical consistency, bringing high extrapolation capacity for changing operational conditions. The key finding is that physical consistency can be linked to convexity, a central concept in optimization. Therefore, convexified ML designs are proposed and the global optimality implies faithfulness to the underlying physics. Chapter 3 handles scenario (ii) and the goal is to identify the event time, type, and locations. The problem is formalized as multi-class classification with special attention to accuracy and speed. Subsequently, Chapter 3 builds an ensemble learning framework to aggregate different ML models for better prediction. Next, to tackle high-volume data quickly, a tensor as the multi-dimensional array is used to store and process data, yielding compact and informative vectors for fast inference. Finally, if no labels exist, Chapter 3 uses physical properties to generate labels for learning. Chapter 4 deals with scenario (iii) and a doable process is to transfer knowledge from similar systems, under the framework of Transfer Learning (TL). Chapter 4 proposes cutting-edge system-level TL by considering the network structure, complex spatial-temporal correlations, and different physical information.

The past few years have witnessed a significant growth of distributed energy resources (DERs) in power systems at the customer level. Such growth challenges the traditional centralized model of conventional synchronous generation, making a transition to a decentralized network with a significant increase of DERs. This decentralized network requires a paradigm change in modeling distribution systems in more detail to maintain the reliability and efficiency while accommodating a high level of DERs. Accurate models of distribution feeders, including the secondary network, loads, and DER components must be developed and validated for system planning and operation and to examine the distribution system performance. In this work, a detailed model of an actual feeder with high penetration of DERs from an electrical utility in Arizona is developed. For the primary circuit, distribution transformers, and cables are modeled. For the secondary circuit, actual conductors to each house, as well as loads and photovoltaic (PV) units at each premise are represented. An automated tool for secondary network topology construction for load feeder topology assignation is developed. The automated tool provides a more accurate feeder topology for power flow calculation purposes. The input data for this tool consists of parcel geographic information system (GIS) delimitation data, and utility secondary feeder topology database. Additionally, a highly automated, novel method to enhance the accuracy of utility distribution feeder models to capture their performance by matching simulation results with corresponding field measurements is presented. The method proposed uses advanced metering infrastructure (AMI) voltage and derived active power measurements at the customer level, data acquisition systems (DAS) measurements at the feeder-head, in conjunction with an AC optimal power flow (ACOPF) to estimate customer active and reactive power consumption over a time horizon, while accounting for unmetered loads. The method proposed estimates both voltage magnitude and angle for each phase at the unbalanced distribution substation. The accuracy of the method developed by comparing the time-series power flow results obtained from the enhancement algorithm with OpenDSS results and with the field measurements available. The proposed approach seamlessly manages the data available from the optimization procedure through the final model verification.

Due to the new and old challenges, modern-day market management systems continue ‎to evolve, including market reformulations, introducing new market products, and ‎proposing new frameworks for integrating distributed energy resources (DERs) into the ‎wholesale markets. Overall, questions is regarding how to reflect these essential changes in ‎the market models (design, reformulation, and coordination frameworks), design market-‎based incentive structures to adequately compensate participants for providing ancillary ‎services, and assess these impacts on market settlements.‎First, this dissertation proposes the concept of securitized-LMP to solve the issue of how ‎market participants should be compensated for providing N-1 reliability services. Then, ‎pricing implications and settlements of three state-of-art market models are compared. The ‎results show that with a more accurate representation of contingencies in the market ‎models, N-1 grid security requirements are originally captured; thereby, the value of service ‎provided by generators is reflected in the prices to achieve grid security.‎ Also, new flexible ramping product (FRP) designs are proposed for different market ‎processes to (i) schedule day-ahead (DA) FRP awards that are more adaptive concerning ‎the real-time (RT) 15-min net load changes, and (ii) address the FRP deployability issue in ‎fifteen-minute market (FMM). The proposed market models performance with enhanced ‎FRP designs is compared against the DA market and FMM models with the existing FRP ‎design through a validation methodology based on California independent system operator ‎‎(ISO) RT operation. The proposed FRP designs lead to less expected final RT operating ‎cost, higher reliability, and fewer RT price spikes.‎ Finally, this dissertation proposes a distribution utility and ISO coordination framework ‎to enable ISO to manage the wholesale market while preemptively not allowing ‎aggregators to cause distribution ‎system (DS) violations. To this end, this coordination ‎framework architecture utilizes the statistical information obtained using different DS ‎conditions and data-mining algorithms to predict the aggregators qualified maximum ‎capacity. A validation phase considering Volt-VAr support provided by distributed PV smart ‎inverters is utilized for evaluate the proposed model performance. The proposed model ‎produces wholesale market awards for aggregators that fall within the DS operational limits ‎and, consequently, will not impose reliable and safety issues for the DS.‎

Although the increasing penetration of electric vehicles (EVs) has reduced the emissionof the greenhouse gas caused by vehicles, it would lead to serious congestion on-road and in charging stations. Strategic coordination of EV charging would benefit the transportation system. However, it is difficult to model a congestion game, which includes choosing charging routes and stations. Furthermore, conventional algorithms cannot balance System Optimization and User Equilibrium, which can cause a huge waste to the whole society. To solve these problems, this paper shows (1) a congestion game setup to optimize and reveal the relationship between EV users, (2) using ε – Nash Equilibrium to reduce the inefficient impact from the self-minded the behavior of the EV users, and (3) finding the relatively optimal solution to approach Pareto-Optimal solution. The proposed method can reduce more total EVs charging time and most EV users’ charging time than existing methods. Numerical simulations demonstrate the advantages of the new method compared to the current methods.

The performance of voltage source inverter (VSI) in terms of output waveform quality, conversion efficiency and common mode noise depends greatly on the pulse width modulation (PWM) method. In this work, a low-loss space vector PWM i.e., 240°-clamped space vector PWM (240CPWM) is proposed to improve the performance of VSIs in electric/hybrid electric vehicles (EV/HEVs) and grid connected photovoltaic (PV) systems. The salient features of 240CPWM include 240° clamping of each phase pole to positive or negative DC bus in a fundamental cycle ensuring that switching losses are reduced by a factor of seven as compared to conventional space vector PWM (CSVPWM) at unity power factor. Zero states are completely eliminated and only two nearest active states are used ensuring that there is no penalty in terms of total harmonic distortion (THD) in line current. The THD of the line current is analyzed using the notion of stator flux ripple and compared with conventional and discontinuous PWM method. Discontinuous PWM methods achieve switching loss reduction at the expense of higher THD while 240CPWM achieves a much greater loss reduction without impacting the THD. The analysis and performance of 240CPWM are validated on a 10 kW two-stage experimental prototype. Common mode voltage (CMV) and leakage current characteristics of 240CPWM are analyzed in detail. It is shown analytically that 240CPWM reduces the CMV and leakage current as compared to other PWM methods while simultaneously reducing the switching loss and THD. Experimental results from a 10-kW hardware prototype conform to the analytical discussions and validate the superior performance of 240CPWM. 240CPWM requires a six-pulse dynamic DC link voltage that introduces low frequency harmonics in DC input current and/or AC line currents that can affect maximum power point tracking, battery life or THD in line current. Four topologies have been proposed to minimize the low frequency harmonics in input and line currents in grid-connected PV system with 240CPWM. In order to achieve further benefits in terms of THD and device stress reduction, 240CPWM is extended to three-level inverters. The performance metrics such as THD and switching loss for 240CPWM are analyzed in three-level inverter.

Voltage Source Inverter (VSI) is an integral component that converts DC voltage to AC voltage suitable for driving the electric motor in Electric Vehicles/Hybrid Electric Vehicles (EVs/HEVs) and integration with electric grid in grid-connected photovoltaic (PV) converter. Performance of VSI is significantly impacted by the type of Pulse Width Modulation (PWM) method used.In this work, a new PWM method called 240° Clamped Space Vector PWM (240CPWM) is studied extensively. 240CPWM method has the major advantages of clamping a phase to the positive or negative rail for 240° in a fundamental period, clamping of two phases simultaneously at any given instant, and use of only active states, completely eliminating the zero states. These characteristics lead to a significant reduction in switching losses of the inverter and lower DC link capacitor current stress as compared to Conventional Space Vector PWM. A unique six pulse dynamically varying DC link voltage is required for 240CPWM instead of constant DC link voltage to maintain sinusoidal output voltage. Voltage mode control of DC-DC stage with Smith predictor is developed for shaping the dynamic DC link voltage that meets the requirements for fast control. Experimental results from a 10 kW hardware prototype with 10 kHz switching frequency validate the superior performance of 240CPWM in EV/HEV traction inverters focusing on loss reduction and DC link capacitor currents. Full load efficiency with the proposed 240CPWM for the DC-AC stage even with conventional Silicon devices exceeds 99%. Performance of 240CPWM is evaluated in three phase grid-connected PV converter. It is verified experimentally that 240CPWM performs well under adverse grid conditions like sag/swell and unbalance in grid voltages, and under a wide range of power factor. Undesired low frequency harmonics in inverter currents are minimized using the Harmonic Compensator that results in Total Harmonic Distortion (THD) of 3.5% with 240CPWM in compliance with grid interconnection standards. A new, combined performance index is proposed to compare the performance of different PWM schemes in terms of switching loss, THD, DC link current stress, Common Mode Voltage and leakage current. 240CPWM achieves the best value for this index among the PWM methods studied.