For the purpose of developing a new highly-sensitive and reliable fiber optical acoustic sensor capable of real-time on-line detection of acoustic emissions in power transformers, this dissertation presents the comprehensive research work on the theory, modeling, design, instrumentation, noise analysis, and performance evaluation of a diaphragm-based optical fiber acoustic (DOFIA) sensor system. The optical interference theory and the diaphragm dynamic vibration analysis form the two foundation stones of the diaphragm-based optical fiber interferomtric acoustic (DOFIA) sensor. Combining these two principles, the pressure sensitivity and frequency response of the acoustic sensor system is analyzed quantitatively, which provides guidance for the practical design for the DOFIA sensor probe and system. To meet all the technical requirements for partial discharge detection, semiconductor process technologies are applied, for the first time to our knowledge, in fabricating the micro-caved diaphragm (MCD) used for the DOFIA sensor probe. The novel controlled thermal bonding method was proposed, designed, and developed to fabricate high performance DOFIA sensor probes with excellent mechanical strength and temperature stability. In addition, the signal processing unit is designed and implemented with high gain, wide band response, and ultra low noise. A systematic noise analysis is also presented to provide a better understanding of the performance limitations of the DOFIA sensor system. Based on the system noise analysis results, optimization measures are proposed to improve the system performance. Extensive experiments, including the field testing in a power transformer, have also been conducted to systematically evaluate the performance of the instrumentation systems and the sensor probes. These results clearly demonstrated the feasibility of the developed DOFIA sensor for the detection of partial discharges inside electrical power transformers, with unique advantages of non-electrically conducting, high sensitivity, high frequency response, and immunity to the electro-magnetic interference (EMI). / Ph. D.

LLC resonant converters have been a popular choice for DC-DC converters due to their high efficiency, high power density, and hold-up capability in power supplies for communication systems, datacenters, consumer electronics, and automobiles. With the rapid development of wide-bandgap devices and novel magnetic materials, the push for higher switching frequencies to achieve higher power densities at lower costs is gaining traction. To demonstrate high efficiency and high power density, the Center for Power Electronics Systems (CPES) at Virginia Tech designed an 800W, 1MHz 400V/12V LLC converter for future datacenters, which could achieve a peak efficiency of 97.6% and a power density of 900 W/in3. However, with the ever-increasing demand for online services, the performance of power delivery must also be simultaneously improved to keep pace with the demand. The focus of this thesis is improving the performance of CPES’ previous 400V/12V LLC converter by investigating different aspects of its design and operation. Ultimately, design guidelines are proposed, and improvements are demonstrated to effectively achieve higher efficiency and higher power density than the previous CPES converter. Multiple aspects of the LLC converter’s design and structure are investigated to further improve its performance, and three main areas are the focus of this thesis. The output-side termination design of the planar transformer is investigated and modeled, and design guidelines for filter capacitor selection are provided for optimal efficiency. Next, the existing shielding technique for matrix transformers, which helps reduce common-mode (CM) noise without compromising on efficiency, is investigated for asymmetry and current-sharing issues, and modifications have been proposed to improve its efficiency. Thirdly, the LLC converter’s switching frequency is optimized to improve its performance over the previous CPES converter. Finally, the hardware results with the proposed improvements are demonstrated, and the converter’s performance is compared with the previous CPES converter as well as other recent proposed solutions. / M.S. / The electricity demand by datacenters has been growing exponentially over the past few decades, especially due to the boom of artificial intelligence in addition to other internet services. This has resulted in a requirement to continually improve the efficiencies of the power delivery from the grid, through the datacenter power architecture, and finally to the loads on the server racks. The overall datacenter power architecture has been improved over time to improve the total efficiency. However, the performance of each stage along the power architecture must be improved to keep in pace with the energy demand. The focus of this thesis is to improve the performance of the 400V/12V DC-DC stage for future datacenters. Previously, the Center for Power Electronics Systems (CPES) at Virginia Tech developed a 1MHz 800W 400V/12V LLC converter with 97.6% peak efficiency and 900W/in3 power density. However, the performance of the converter must be further improved to stay ahead of the competition and keep in pace with the increasing energy demand. Multiple aspects of the LLC converter’s design and structure are investigated to further improve its performance, and three main areas are the focus of this thesis. Firstly, the high-frequency termination design, or how different components are interconnected and arranged, is studied, and a capacitance selection guideline is proposed to maximize the efficiency. Next, the existing shielding technique for matrix transformers, which helps reduce common-mode (CM) noise without compromising on efficiency, is investigated for asymmetry and current-sharing issues, and modifications have been proposed to improve its efficiency. Thirdly, the LLC converter’s switching frequency is optimized to improve its performance over the previous CPES converter. Finally, the hardware results with the proposed improvements are demonstrated, and the converter’s performance is compared with the previous CPES converter as well as other recent proposed solutions.

The intermediate bus architecture employing the 48V bus converter is one of the most popular power architecture. 48V to 12V bus converter has wide applications in telecommunications, networks, aerospace, and military, etc. However, today's state of the art products has low power rating or power density and becomes difficult to satisfy the demand of increasing power of the loads. To improve the current design, a GaN (Gallium Nitride) based two-stage solution is proposed for the bus converter. The first stage Buck converter regulates the 40V to 60V variable input to a fixed 36V bus voltage. The second stage LLC converter convert the 36V to 12V by a 3:1 transformer. The whole solution achieves the fixed frequency control. The thesis focus on the detail design and optimization of LLC converter, especially its transformer. To have high density and high efficiency, the transformer design becomes critical at MHz frequency. The matrix transformer concept is applied and a merged winding structure is used for flux cancellation, which effectively reduces the AC winding losses. A new fully interleaved termination and via design is proposed. It achieves significant reduction in loss and leakage flux. In addition, to study the current sharing of parallel winding layers, a 1-D analytic model is proposed and a symmetrical winding layer scheme is used to balance the current distribution. The hardware is built and tested. The proposed two-stage converter achieves the best performance compared to the current market. / Master of Science / Intermediate bus architecture (IBA) has wide applications in telecommunication, server and computing, and military power supplies. The intermediate bus converter (IBC) is the key stage in the IBA, where the DC bus voltage from the front-end power supply is converted to a lower intermediate bus voltage. Traditional IBC suffers from bulky magnetic components including inductors and transformers. This work illustrates the design and implementation of a two-stage IBC, where the first-stage Buck converter will provide regulation and the second stage LLC converter will provide isolation. Thanks to the soft-switching capability of LLC, the magnetic volume can be significantly reduced by raising the switching frequency of the converter. Therefore, planar magnetics can be used and placed directly inside of the printing circuit board (PCB), which allows for higher power densities and easy manufacturing of the magnetics and overall converter. However, as the frequency goes higher, the AC losses of the transformer caused by the eddy current, skin effect, and proximity effect become dominant. As a result, high-frequency transformer design becomes the key for the converter design. First, matrix transformer concept is applied to distribute the high current and reduce the conduction loss. Second, a novel merged winding structure is proposed for better transformer winding interleaving. Third, a new terminal structure of the transformer is proposed. Finally, the current sharing between parallel windings are modeled and studied. All the efforts result in great loss reduction. The prototype were verified and compared to the current converters that are on the market in the 48V – 12V area of IBCs.

For the purpose of this study, data for low voltage distribution transformer loading in small communities in Maryland was collected from a local electric utility company. Specifically, analysis was done on three distribution transformers on their system. Each of these transformers serves at least one electric vehicle (EV) owner. Of the three transformers analyzed, Transformer 2 serves eight residential homes and has the highest risk of experiencing an overload if all customers purchase at least one EV. Transformer 2 has a nameplate rating of 25kVA (22.5kW assuming a 0.9 power factor). With one EV owner, Transformer 2 has a peak load of 46.82kW during the study period between August 4 and August 17, 2013. When seven additional EVs of different types were added in a simulated scenario, the peak load for Transformer 2 increased from 46.82kW to 89.76kW, which is outside the transformer thermal limit. With the implementation of TOU pricing, the peak load was reduced to 56.71kW from 89.76kW. By implementing a combination of TOU pricing and appliance cycling through demand side management (DSM), the peak load was further reduced to 52.27kW. / Master of Science

Magnetoelectric (ME) effect is defined as the change in dielectric polarization (P) of a material under an applied magnetic field (H) or an induced magnetization (M) under an external electric field (E). ME materials have attracted number of investigators due to their potential for improving applications such as magnetic field sensors, filters, transformers, memory devices and energy harvesters. It has been shown both experimentally and theoretically that the composite structures consisting of piezoelectric and magnetostrictive phases possess stronger ME coupling in comparison to that of single phase materials. Giant magnetoelectric effect has been reported in variety of composites consisting of bulk-sized ME composites and thin film ME nanostructures. In this dissertation, novel ME composite systems are proposed, synthesized and characterized in both bulk and thin films to address the existing challenges in meeting the needs of practical applications. Two applications were the focused upon in this study, tunable transformer and dual phase energy harvester, where requirements can be summarized as: high ME coefficient under both on-resonance and off-resonance conditions, broad bandwidth, and low applied DC bias. In the first chapter, three challenges related to the conventional ME behavior in bulk ME composites have been addressed (1) The optimized ME coefficient can be achieved without external DC magnetic field by using a self-biased ME composite with a homogenous magnetostrictive material. The mechanism of such effect and its tunability are studied; (2) A near-flat ME response regardless of external magnetic field is obtained in a self-biased ME composite with geometry gradient structure; (3) By optimizing interfacial coupling with co-firing techniques, the ME coefficient can be dramatically enhanced. Theses co-fired ME laminates not only exhibit high coupling coefficient due to direct bonding, but also illustrate a self-biased effect due to the built-in stress during co-sintering process. These results present significant advancement toward the development of multifunctional ME devices since it eliminates the need for DC bias, expands the working bandwidth and enhances the ME voltage coefficient. Next, magnetoelectric nanocomposites were developed for understanding the nature of the growth of anisotropic thin film structures. In this chapter following aspects were addressed: (1) Controlled growth of nanostructures with well-defined morphology was obtained. Microstructure and surface morphology evolution of the piezoelectric BaTiO3 films was systematically analyzed. A growth model was proposed by considering the anisotropy of surface energy and the formation of twin lamellae structure within the frame work of Structure Zone Model (SZM) and Dynamic Scaling Theory (DST). In parallel to BaTiO3 films, well-ordered nanocomposite arrays [Pb1.1(Zr0.6Ti0.4)O3/CoFe2O4] with controlled grain orientation were developed and investigated by a novel hybrid deposition method. The influence of the pre-deposited template film orientation on the growth of ME composite array was studied. (2) PZT/CFO/PZT thick composite film and BTO/CFO thin film were synthesized using sol-gel deposition (SGD) and pulsed laser deposition (PLD) techniques, respectively. The HRTEM analysis revealed local microstructure at the interface of consecutive constituents. The interfacial property variation of these films was found to affect the coupling coefficient of corresponding ME nanocomposites. Subsequently, a novel complex three-dimensional ME composite with highly anisotropic structure was developed using a hybrid synthesis method. The influence of growth condition on the microstructure and property of the grown complex composites was studied. The film with highly anisotropic structure was found to possess tailored ferroelectric response indicating the promise of this synthesis method and microstructure. Based on the laminated ME composites, three types of ME tunable transformer designs were designed and fabricated. The goal was to develop a novel ME transformer with tunable performance (voltage gain and/or working resonance frequency) under applied DC magnetic field. Conventional ME transformers need either winding coil or large external magnetic field to achieve the tunable feature. Considering the high ME coupling of ME laminate, two ME transformers were developed by epoxy bonding Metglas with transversely/longitudinally poled piezoelectric ceramic transformer. The influence of different operation modes toward magnetoelectric tunability was analyzed. In addressing the concern of the epoxy bonding interface, a co-fired ME transformer with unique piezoelectric transformer/magnetostrictive layer/piezoelectric transformer trilayer structure was designed. The design and development strategy of thin film ME transformer was discussed to illustrate the potential for ME transformer miniaturization and on-chip integration. Lastly, motivated by the increasing demand of energy harvesting (EH) systems to support self-powered sensor nodes in structural health monitoring system, a magnetoelectric composite based energy harvester was developed. The development and design concept of the magnetoelectric energy harvester was systematically discussed. In particular, the first dual-phase self-biased ME energy harvester was designed which can simultaneously harness both vibration and stray magnetic field (Hac) in the absence of DC magnetic field. Strain distribution of the EH was simulated using the finite element model (FEM) at the first three resonance frequencies. Additionally, the potential of transferring this simple EH structure into MEMS scalable components was mentioned. These results provide significant advancement toward high energy density multimode energy harvesting system. / Ph. D.

In the early 1990's, with the boom of the Internet and the advancements in telecommunications, the demand for high-speed communications systems has reached every corner of the world in forms such as, phone exchanges, the internet servers, routers, and all other types of telecommunication systems. These communication systems demand more data computing, storage, and retrieval capabilities at higher speeds, these demands place a great strain on the power system. To lessen this strain, the existing power architecture must be optimized. With the arrival of the age of high speed and power hungry microprocessors, the point of load converter has become a necessity. The power delivery architecture has changed from a centralized distribution box delivering an entire system's power to a distributed architecture, in which a common DC bus voltage is distributed and further converted down at the point of load. Two common distributed bus voltages are 12 V for desktop computers and 48 V for telecommunications server applications. As industry strives to design more functionality into each circuit or motherboard, the area available for the point of load converter is continually decreasing. To meet industries demands of more power in smaller sizes power supply designers must increase the converter's switching frequencies. Unfortunately, as the converter switching frequency increases the efficiency is compromised. In particular, the switching, gate drive and body diode related losses proportionally increase with the switching frequency. This thesis introduces a loss saving self-driven method to drive the secondary side synchronous rectifiers. The loss saving self-driven method introduces two additional transformers that increase the overall footprint of the converter. Also, this thesis proposes a new magnetic integration method to eliminate the need for the two additional gate driver magnetic cores by allowing three discrete power signals to pass through one single magnetic structure. The magnetic integration reduces the overall converter footprint. / Master of Science

Neural networks typically struggle with reasoning tasks on out of domain data, something that humans can more easily adapt to. Humans come with prior knowledge of concepts and can segment their environment into building blocks (such as objects) that allow them to reason effectively in unfamiliar situations. Using this intuition, we train a network that utilizes fixed embeddings from the CLIP (Contrastive Language--Image Pre-training) model to do a simple task that the original CLIP model struggles with. The network learns concepts (such as "collide" and "avoid") in a supervised source domain in such a way that the network can adapt and identify similar concepts in a target domain with never-before-seen objects. Without any training in the target domain, we show a 11% accuracy improvement in recognizing concepts compared to the baseline zero-shot CLIP model. When provided with a few labels, this accuracy gap widens to 20%.

Multi-pulse transformer/rectifier systems are becoming increasingly popular in power distribution systems. These topologies can be found in aircraft power systems, motor drives, and other applications that require low total harmonic distortion (THD) of the input line current. This increase in the use of multi-pulse transformer topologies has led to the need to study large systems composed of said units and their interactions within the system. There is also an interest in developing small-signal models so that stability issues can be studied. This thesis presents a procedure for the average model of multi-pulse transformer/rectifier topologies. The dq rotating reference frame was used to develop the average model and parameter estimation is incorporated through the use of polynomial fits. The average model is composed of nonlinear dependent sources and linear passive components. A direct benefit from this approach is a reduction in simulation time by two orders of magnitude. The average model concept demonstrates that it accurately predicts the dynamics of the system being studied. In particular, two specific topologies are studied, the 12-pulse hexagon transformer/rectifier (hex t/r) and the 18-pulse autotransformer rectifier unit (ATRU). In both cases, detailed switching model results are used to verify the operation of the average model. In the case of the hex t/r, the average model is further validated with experimental data from an 11 kVA prototype. The hex t/r output impedance, obtained from the linearized average model, has also been verified experimentally. / Master of Science

Over the recent years many people have been trying to reduce the size and weight of magnetic components and thus the overall system [ 19 ]. One attempt at this is to increase the switching frequency of the system. However, this attempt has its limitations due to increased device switching losses. Device limitations usually confine this frequency to lower value than is desired. An effective approach, reducing the size and weight is to use the planar magnetics for possible integration with the power circuit and thus eliminating the associated interconnections. Planar magnetics uses the printed circuit board as the windings. This will allow the magnetic component to be implemented into the circuit. The integration of the magnetic components and power circuit will decrease the number of connections, reduce the height, and ensure the parasitic repeatability. Having external connections can cause problems in the system. In this case the system must carry a large amount of current. The connections can cause heating from resistance and inductance of the connection. The planar approach also will decrease the height of the system. This is because the planar magnetic cores have a higher surface area with a decreased height. This can reduce the height of the system by 25 %- 50 % [ 19 ]. The parasitic repeatability is also a very important factor. In many cases the typology relies on the parasitic elements for energy storage. Since, the parasitic elements are mainly a result from the geometry of the system; and the planar system has the windings made from the printed circuit board, the parasitic elements will be very consistent through the manufacturing process. For topologies that rely on the parasitic elements for soft switching, the planar design can incorporate parasitic elements with the leakage components for the soft-switching requirement. This thesis redefines the conventional term of leakage inductance as the sum of a set of lumped parasitic inductances and the transformer leakage inductance for the integrated planar magnetics and inverter power circuitry. For the conventional non-integrated transformer, either planar or non-planar, the leakage inductance is defined between two terminals of the transformer. However, for the integrated planar magnetics, the new lumped parasitic and leakage inductance should include the inverter switch and dc bus interconnections. The transformer was first designed using a closed-form solution for a known geometry with different copper thickness. The calculated leakage inductance was then verified with finite element analysis and the impedance analyzer measurement. It was found that the theoretical calculation and the finite element analysis results agreed very well, but the measurement was more than one order of magnitude higher. This prompted the study of interconnect parasitics. With geometrical structure and proper termination and lumping, a set of parasitic inductances were defined, and the results were verified with measurements of both impedance analyzer and phase-shifted modulated full-bridge inverter testing. In addition to parasitic inductance analysis, the flux distribution and associated thermal performance of the planar structure were also studied with finite element analysis. The resulting plots of flux distribution and temperature profile indicate the key locations of mechanical mounting and heat sinking. Overall the thesis covers essential design considerations in electrical, mechanical, and thermal aspects for the planar magnetics integration. / Master of Science

This thesis delves into the advancement of 5G portable test-nodes in wireless communication systems with cognitive radio capabilities, specifically addressing the critical need for dynamic spectrum sensing and awareness at the radio receiver through AI-driven automatic modulation classification. Our methodology is centered around the transformer encoder architecture incorporating a multi-head self-attention mechanism. We train our architecture extensively across a diverse range of signal-to-noise ratios (SNRs) from the RadioML 2018.01A dataset. We introduce a novel transformer-based architecture with a gated mechanism, designed as a runtime re-configurable automatic modulation classification framework, which demonstrates enhanced performance with low SNR RF signals during evaluation, an area where conventional methods have shown limitations, as corroborated by existing research. Our innovative single-model framework employs distinct weight sets, activated by varying SNR levels, to enable a gating mechanism for more accurate modulation classification. This advancement in automatic modulation classification marks a crucial step toward the evolution of smarter communication systems. / Master of Science / This thesis delves into the advancement of wireless communication systems, particularly in developing portable devices capable of effectively detecting and analyzing radio signals with cognitive radio capabilities. Central to our research is leveraging artificial intelligence (AI) for automatic modulation classification, a method to identify signal modulation types. We utilize a transformer-based AI model trained on the RadioML 2018.01A dataset. Our training approach is particularly effective when evaluating low-quality signals using a gating mechanism based on signal-to-noise ratios, an area previously considered challenging in existing research. This work marks a significant advancement in creating more intelligent and responsive wireless communication systems.