The high uncertainty of renewables introduces more dynamics to power systems. The conventional way of monitoring and controlling power systems is no longer reliable. New strategies are needed to ensure the stability and reliability of power systems. This work aims to assess the use of machine learning methods in analyzing data from renewable integrated power systems to aid the decisionmaking of electricity market participants. Specifically, the work studies the cases of electricity price forecast, solar panel detection, and how to constrain the machine learning methods to obey domain knowledge.Chapter 2 proposes to diversify the data source to ensure a more accurate electricity price forecast. Specifically, the proposed two-stage method, namely the rerouted method, learns two types of mapping rules: the mapping between the historical wind power and the historical price and the forecasting rule for wind generation. Based on the two rules, we forecast the price via the forecasted generation and the learned mapping between power and price. The massive numerical comparison gives guidance for choosing proper machine learning methods and proves the effectiveness of the proposed method. Chapter 3 proposes to integrate advanced data compression techniques into machine learning algorithms to either improve the predicting accuracy or accelerate the computation speed. New semi-supervised learning and one-class classification methods are proposed based on autoencoders to compress the data while refining the nonlinear data representation of human behavior and solar behavior. The numerical results show robust detection accuracy, laying down the foundation for managing distributed energy resources in distribution grids. Guidance is also provided to determine the proper machine learning methods for the solar detection problem. Chapter 4 proposes to integrate different types of domain knowledge-based constraints into basic neural networks to guide the model selection and enhance interpretability. A hybrid model is proposed to penalize derivatives and alter the structure to improve the performance of a neural network. We verify the performance improvement of introducing prior knowledge-based constraints on both synthetic and real data sets.

Modular multilevel converters (MMCs) have become an attractive technology for high power applications. One of the main challenges associated with control and operation of the MMC-based systems is to smoothly precharge submodule (SM) capacitors to the nominal voltage during the startup process. The existing closed-loop methods require additional effort to analyze the small-signal model of MMC and tune control parameters. The existing open-loop methods require auxiliary voltage sources to charge SM capacitors, which add to the system complexity and cost. A generalized precharging strategy is proposed in this thesis.

For large-scale MMC-embedded power systems, it is required to investigate dynamic performance, fault characteristics, and stability. Modeling of the MMC is one of the challenges associated with the study of large-scale MMC-based power systems. The existing models of MMC did not consider the various configurations of SMs and different operating conditions. An improved equivalent circuit model is proposed in this thesis.

The solid state transformer (SST) has been investigated for the distribution systems to reduce the volume and weight of power transformer. Recently, the MMC is employed into the SST due to its salient features. For design and control of the MMC-based SST, its operational principles are comprehensively analyzed. Based on the analysis, its mathematical model is developed for evaluating steady-state performances. For optimal design of the MMC-based SST, the mathematical model is modified by considering circuit parameters.

One of the challenges of the MMC-based SST is the balancing of capacitor voltages. The performances of various voltage balancing algorithms and different modulation methods have not been comprehensively evaluated. In this thesis, the performances of different voltage-balancing algorithms and modulation methods are analyzed and evaluated. Based on the analysis, two improved voltage-balancing algorithms are proposed in this thesis.

For design of the MMC-based SST, existing references only focus on optimal design of medium-frequency transformer (MFT). In this thesis, an optimal design procedure is developed for the MMC under medium-frequency operation based on the mathematical model of the MMC-based SST. The design performance of MMC is comprehensively evaluated based on free system parameters.

Nearly all solar photovoltaic (PV) systems are designed with maximum power point tracking (MPPT) functionality to maximize the utilization of available power from the PV array throughout the day. In conventional PV systems, the MPPT function is handled by a power electronic device, like a DC-AC inverter. However, given that most PV systems are designed to be grid-connected, there are several challenges for designing PV systems for DC-powered applications and off-grid applications. The first challenge is that all power electronic devices introduce some degree of power loss. Beyond the cost of the lost power, the upfront cost of power electronics also increases with the required power rating. Second, there are very few commercially available options for DC-DC converters that include MPPT functionality, and nearly all PV inverters are designed as “grid-following” devices, as opposed to “grid-forming” devices, meaning they cannot be used in off-grid applications.

To address the challenges of designing PV systems for high-power DC and off-grid applications, a load-managing photovoltaic (LMPV) system topology has been proposed. Instead of using power electronics, the LMPV system performs maximum power point tracking through load management. By implementing a load-management approach, the upfront costs and the power losses associated with the power electronics are avoided, both of which improve the economic viability of the PV system. This work introduces the concept of an LMPV system, provides in-depth analyses through both simulation and experimental validation, and explores several potential applications of the system, such as solar-powered commercial-scale electrolyzers for the production of hydrogen fuel or the production and purification of raw materials like caustic soda, copper, and zinc.

The objectives of this research project were to develop a model of real power demand from a dc fast charging station both with and without an integrated battery energy storage system (BESS). An optimal deterministic control strategy was devel-oped to perform load-shaping under various scenarios with various load-shaping goals in mind to establish the value for BESS’s with various power and energy capacities.

To achieve these objectives, first a statistical model of electric vehicle drivers’ charging behaviors (home charging and dc fast charging) was constructed and simu-lated according to empirical charging data and several key findings about people’s charging habits in the literature.

Data of private vehicles’ driving records was extracted from the National Household Travel Survey (NHTS), derived 42 statistical distributions that mathe-matically modeled people’s driving behaviors. From this start, two algorithms were developed to simulate driver behavior: one using a database sampling method (DSM) and another using probability distribution sampling method (PDSM) to simulate the electric vehicles’ driving cycles. Both methods used data and statistical distributions derived from NHTS. Next, a model of the EV drivers’ charging behavior was incor-porated into the simulation of the electric vehicles’ driving cycles, and then the ve-hicles’ charging behaviors were simulated. From these simulations, one can forecast the real-power demand of a typical dc fast charging station with six dc 50 kW fast chargers serving a population of 700 EVs. (The ratio of six dc fast chargers to 700 EVs was selected based on the current value of this ratio in the US.) Next, a BESS was integrated into the dc fast charging station demand model and the size and charging behavior was optimized to account for different criteria which were based on the goals of the different potential owners: SRP or a third-party owner. It was established when a BESS would become economically feasible using a simplified economic model.

It was observed that the real-power demand shape is a function of the size of the BESS and the owner’s objective, i.e., flattening the demand curve or minimizing the cost of electricity.

Due to the rapid penetration of solar power systems in residential areas, there has

been a dramatic increase in bidirectional power flow. Such a phenomenon of bidirectional

power flow creates a need to know where Photovoltaic (PV) systems are

located, what their quantity is, and how much they generate. However, significant

challenges exist for accurate solar panel detection, capacity quantification,

and generation estimation by employing existing methods, because of the limited

labeled ground truth and relatively poor performance for direct supervised learning.

To mitigate these issue, this thesis revolutionizes key learning concepts to (1)

largely increase the volume of training data set and expand the labelled data set by

creating highly realistic solar panel images, (2) boost detection and quantification

learning through physical knowledge and (3) greatly enhance the generation estimation

capability by utilizing effective features and neighboring generation patterns.

These techniques not only reshape the machine learning methods in the GIS

domain but also provides a highly accurate solution to gain a better understanding

of distribution networks with high PV penetration. The numerical

validation and performance evaluation establishes the high accuracy and scalability

of the proposed methodologies on the existing solar power systems in the

Southwest region of the United States of America. The distribution and transmission

networks both have primitive control methodologies, but now is the high time

to work out intelligent control schemes based on reinforcement learning and show

that they can not only perform well but also have the ability to adapt to the changing

environments. This thesis proposes a sequence task-based learning method to

create an agent that can learn to come up with the best action set that can overcome

the issues of transient over-voltage.

This dissertation covers three primary topics and relates them in context. High frequency transformer design, microgrid modeling and control, and converter design as it pertains to the other topics are each investigated, establishing a summary of the state-of-the-art at the intersection of the three as a baseline. The culminating work produced by the confluence of these topics is a novel modular solid-state transformer (SST) design, featuring an array of dual active bridge (DAB) converters, each of which contains an optimized high-frequency transformer, and an array of grid-forming inverters (GFI) suitable for centralized control in a microgrid environment. While no hardware was produced for this design, detailed modeling and simulation has been completed, and results are contextualized by rigorous analysis and comparison with results from published literature. The main contributions to each topic are best presented by topic area. For transformers, contributions include collation and presentation of the best-known methods of minimum loss high-frequency transformer design and analysis, descriptions of the implementation of these methods into a unified design script as well as access to an example of such a script, and the derivation and presentation of novel tools for analysis of multi-winding and multi-frequency transformers. For microgrid modeling and control, contributions include the modeling and simulation validation of the GFI and SST designs via state space modeling in a multi-scale simulation framework, as well as demonstration of stable and effective participation of these models in a centralized control scheme under phase imbalance. For converters, the SST design, analysis, and simulation are the primary contributions, though several novel derivations and analysis tools are also presented for the asymmetric half bridge and DAB.

As the world becomes more electronic, power electronics designers have continuously designed more efficient converters. However, with the rising number of nonlinear loads (i.e. electronics) attached to the grid, power quality concerns, and emerging legislation, converters that intake alternating current (AC) and output direct current (DC) known as rectifiers are increasingly implementing power factor correction (PFC) by controlling the input current. For a properly designed PFC-stage inductor, the major design goals include exceeding minimum inductance, remaining below the saturation flux density, high power density, and high efficiency. In meeting these goals, loss calculation is critical in evaluating designs. This input current from PFC circuitry leads to a DC bias through the filter inductor that makes accurate core loss estimation exceedingly difficult as most modern loss estimation techniques neglect the effects of a DC bias. This thesis explores prior loss estimation and design methods, investigates finite element analysis (FEA) design tools, and builds a magnetics test bed setup to empirically determine a magnetic core’s loss under any electrical excitation. In the end, the magnetics test bed hardware results are compared and future work needed to improve the test bed is outlined.

The following report details the motivation, design, analysis, simulation and hardware implementation of a DC/DC converter in EV drivetrain architectures. The primary objective of the project was to improve overall system efficiency in an EV drivetrain. The methodology employed to this end required a variable or flexible DC-Link voltage at the input of the inverter stage. Amongst the several advantages associated with such a system are the independent optimization of the battery stack and the inverter over a wide range of motor operating conditions. The incorporation of a DC/DC converter into the drivetrain helps lower system losses but since it is an additional component, a number of considerations need to be made during its design. These include stringent requirements on power density, converter efficiency and reliability.

These targets for the converter are met through a number of different ways. The switches used are Silicon Carbide FETs. These are wide band gap (WBG) devices that can operate at high frequencies and temperatures. Since they allow for high frequency operation, a switching frequency of 250 khz is proposed and implemented. This helps with power density by reducing the size of passive components. High efficiencies are made possible by using a simple soft switching technique by augmenting the DC/DC converter with an auxiliary branch to enable zero voltage transition.

The efficacy of the approach is tested through simulation and hardware implementation of two different prototypes. The Gen-I prototype was a single soft switched synchronous boost converter rated at 2.5kw. Both the motoring mode and regenerative modes of operation (Boost and Buck) were hardware tested for over 2kw and efficiency results of over 98.15% were achieved. The Gen-II prototype and the main focus of this work is an interleaved soft switched synchronous boost converter. This converter has been implemented in hardware as well and has been tested at 6.7kw and an efficiency of over 98% has been achieved in the boost mode of operation.

Over the years, the growing penetration of renewable energy into the electricity market has resulted in a significant change in the electricity market price. This change makes the existing forecasting method prone to error, decreasing the economic benefits. Hence, more precise forecasting methods need to be developed. This paper starts with a survey and benchmark of existing machine learning approaches for forecasting the real-time market (RTM) price. While these methods provide sufficient modeling capability via supervised learning, their accuracy is still limited due to the single data source, e.g., historical price information only. In this paper, a novel two-stage supervised learning approach is proposed by diversifying the data sources such as highly correlated power data. This idea is inspired by the recent load forecasting methods that have shown extremely well performances. Specifically, the proposed two-stage method, namely the rerouted method, learns two types of mapping rules. The first one is the mapping between the historical wind power and the historical price. The second is the forecasting rule for wind generation. Based on the two rules, we forecast the price via the forecasted generation and the first learned mapping between power and price. Additionally, we observed that it is not the more training data the better, leading to our validation steps to quantify the best training intervals for different datasets. We conduct comparisons of numerical results between existing methods and the proposed methods based on datasets from the Electric Reliability Council of Texas (ERCOT). For each machine learning step, we examine different learning methods, such as polynomial regression, support vector regression, neural network, and deep neural network. The results show that the proposed method is significantly better than existing approaches when renewables are involved.

Switching surges are a common type of phenomenon that occur on any sort of power system network. These are more pronounced on long transmission lines and in high voltage substations. The problem with switching surges is encountered when a lot of power is transmitted across a transmission line
etwork, typically from a concentrated generation node to a concentrated load. The problem becomes significantly worse when the transmission line is long and when the voltage levels are high, typically above 400 kV. These overvoltage transients occur following any type of switching action such as breaker operation, fault occurrence/clearance and energization, and they pose a very real danger to weakly interconnected systems. At EHV levels, the insulation coordination of such lines is mainly dictated by the peak level of switching surges, the most dangerous of which include three phase line energization and single-phase reclosing. Switching surges can depend on a number of independent and inter-dependent factors like voltage level, line length, tower construction, location along the line, and presence of other equipment like shunt/series reactors and capacitors.

This project discusses the approaches taken and methods applied to observe and tackle the problems associated with switching surges on a long transmission line. A detailed discussion pertaining to different aspects of switching surges and their effects is presented with results from various studies published in IEEE journals and conference papers. Then a series of simulations are presented to determine an arrangement of substation equipment with respect to incoming transmission lines; that correspond to the lowest surge levels at that substation.