Master of Science, Miami University, 2020, Mechanical and Manufacturing Engineering

Design and implementation of active aeroelastic control (AAC) involves understanding the dynamic characteristics and requirements of actuators and their complex aeroservoelastic interaction for a given flight condition. The performance of these actuators depends on their governing dynamics defining their actuation mechanism and various actuator parameters. In order to realize AAC for a given flight envelop, during its operation, power consumption of the actuators needs to be estimated and the corresponding wasted heat generated needs to be managed. In this research, servo-actuators for driving the control surfaces are integrated with aeroelastic models for feedback control design using the receptance method. Their performance for active flutter suppression is evaluated for a given flight envelope in conjunction with the computation of corresponding power consumption and wasted heat generation. The performance of the actuator for various controller designs (full and partial pole assignment and output feedback) to achieve flutter boundary extension is demonstrated with numerical examples associated with a simplified 2D wing model and a full-scale 3D fighter wing model. In order to achieve minimum power requirements by the actuators, some optimization studies are also presented. This research and solution strategy can be instrumental in designing future AAC with different types of actuators.

Committee: Kumar Singh Dr. (Advisor); Raymond Kolonay Dr. (Committee Member); James Chagdes Dr. (Committee Member) Subjects: Aerospace Engineering; Mechanical Engineering

Doctor of Philosophy, The Ohio State University, 2017, Electrical and Computer Engineering

The power system is provisioned to evolve into a smart grid that is greener, safer, and more efficient. DC microgrid, a new form of distribution system and an emerging electrical network on ships/airplanes/electronic devices, has risen to prominence as an important building block of the future grid and many other applications. With proper operation and coordination, DC microgrids can exploit the flexibility in generation as well as consumption units, which have been standing unresponsive for decades, to approach a more robust and efficient grid such that every component in a power grid can reach its full potential to vibrantly participate in grid services. This dissertation presents systematic approaches to solve DC microgrid control and optimization problems that are usually marked by challenges like uncertainty, nonlinearity, tractability, and structural constraint issues. First, it is well known that when a DC microgrid is operated in island mode, the stability critical power balance is shadowed by uncertain and volatile generation and consumption. We propose a robust stability framework containing a set of sufficient conditions to provide provable stability guarantee for such systems. We then further investigate into robust control design to improve the performance of the system. In view of the physical communication structures that commonly exist in a microgrid, decentralized/distributed controllers are recognized to be more applicable in practice for their limited reliance on information transmissions. Nevertheless, with the communication structural constraints, the decentralized/distributed control design problem is NP-hard in general, and an ill-designed controller may as well render an originally operative system unstable. We propose an algorithm to design a structurally constrained controller in such a way that it can guarantee a design direction with provable improving performance. Second, for DC microgrids that are in grid-connected mode, the (open full item for complete abstract)

Committee: Wei Zhang (Advisor); Giorgio Rizzoni (Advisor); Antonio Conejo (Committee Member); Mahesh Illindala (Committee Member); Andrej Rotter (Other) Subjects: Electrical Engineering; Energy; Engineering; Operations Research

Doctor of Engineering, Cleveland State University, 2015, Washkewicz College of Engineering

In this dissertation a three-bus, radial distribution system is proposed to be used as a benchmark for the study of power quality problems (PQPs) and their compensation. To mitigate PQPs in distribution systems, Custom Power Systems (CUPS) devices such as Distribution Static Compensator (D-STATCOM), Dynamic Voltage Restorer (DVR), and Unified Power Quality Conditioner (UPQC) are required. An increasingly popular and practical control technique based on Active Disturbance Rejection Control (ADRC), which is simple and has good disturbance rejection capabilities, is implemented to control CUPS devices. Its performance is compared with the highly dominating Proportional-Integral-Derivative (PID) controller. Modelling of the DVR and the D-STATCOM, together with their respective ADRC and PID controllers, were developed under the MATLAB /Simulink© environment. Simulation results, which included voltage sag/swell, voltage imbalance and current imbalance, proved that the ADRC controllers are attractive for their use with CUPS devices to mitigate the PQPs and can be considered as effective replacements for the PID controller.

Committee: Eugenio F Villaseca PhD (Committee Chair); Alexander Charles PhD (Committee Member); Zhiqiang Gao PhD (Committee Member); Dan Simon PhD (Committee Member); Miron Kaufman PhD (Committee Member); Morinec Allen G. PhD (Committee Member) Subjects: Electrical Engineering; Engineering

Master of Science in Engineering (MSEgr), Wright State University, 2011, Electrical Engineering

Current-mode control is the most popular scheme used for the operation of SMPS (Switch Mode Power Supplies). Current-mode control, also known as current-programmed mode or current-injected control is a multi-loop control scheme that has an inner loop and an outer voltage loop. The current loop controls the inductor peak current while the voltage loop controls the output voltage. The inner loop follows a set program by the outer loop. Some of the most popular small-signal models that predict the small-signal characteristics of current-mode control scheme have been analyzed and compared in this thesis. A PWM dc-dc buck converter in CCM(Continuous Conduction Mode) has been chosen to explain the phenomenon of current-mode control in all these models. Small-signal characteristics are generated in MATLAB using the simplified analytical transfer functions. Some of the important small-signal characteristics include the current loop gain, control-to-output gain with the current-loop closed and outer loop open, audio susceptibility, and output impedance. The two most important models in consideration are: 1) Continuous-Time Model and 2) Peak Current-Mode control Model. Despite the fact that both these models predict the instability of current-mode control at a duty ratio of 0.5, these models differ significantly in deriving the expression for the sampling gain. As a result, their small-signal characteristics differ over a wide frequency range. Also, a very less explored average current mode control is compared with the peak-current mode control based on the similar small-signal characteristics.

Committee: Marian Kazimierczuk PhD (Advisor); Marian Kazimierczuk PhD (Committee Chair); Saiyu Ren PhD (Committee Member); Zhang Xiaodang PhD (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy (Ph.D.), University of Dayton, 2019, Electrical and Computer Engineering

A method is presented to incorporate physics-based controllability assessment in an aircraft Multi-disciplinary Analysis and Design Optimization environment with a target fidelity representing the traditional preliminary aircraft design phase. This method was designed with specific intended application to innovative vehicle concepts such as the Efficient Supersonic Air Vehicle, a tailless fighter-type aircraft which requires the use of innovative control effectors to achieve yaw control requirements. Typically, the layout of an aircraft is determined primarily through empirical design methods with minimal physical evaluation influencing the shape. As a result, the evaluation of new technologies such as these innovative control effectors in the past has been limited to placement and testing of them within existing free real estate on an otherwise complete vehicle design. The hypothesis of this dissertation is that inclusion of such technology in earliest stages of the design process has a greater chance of leading to optimal benefit and potentially a closed design for a tailless fighter-type aircraft. However, incorporation of technology that does not have a strong statistical basis through prior work requires some form of physical analysis to be performed in the design iteration. An aerodynamic study was performed to determine the optimal combination of fidelity and computation time for analyzing these types of configurations for the controls analysis in the MADO environment, resulting in the use of a multi-fidelity approach to aerodynamic analysis. This approach in turn requires a multi-fidelity, parameterized geometric model of the aircraft with automated generation of analysis mesh. In traditional aircraft design, the disciplines involved are isolated from each other in a linear manner such that one finishes prior to another beginning. Multidisciplinary approaches attempt to merge these. However, in open literature the fidelity level of vario (open full item for complete abstract)

Committee: Raúl Ordóñez Ph.D. (Advisor); Raymond Kolonay Ph.D. (Committee Member); Eric Balster Ph.D. (Committee Member); Keigo Hirakawa Ph.D. (Committee Member) Subjects: Aerospace Engineering; Engineering

Doctor of Philosophy, The Ohio State University, 2018, Electrical and Computer Engineering

Microcontroller Units (MCUs) are central and essential to many consumer electronic and industrial applications, including communication systems, automotive, and Internet of Things (IoT). Since, these MCUs can be used in various applications with different operating conditions, designing the internal power supply of such MCUs is quite challenging. For example, in some applications the MCUs could be powered from a Li-ion battery while in other application it could be powered from on-board regulator, or even an AC-to-DC adapter. This indeed requires the internal power supply of such MCUs to handle a very wide range of input voltages. In addition, these MCUs typically contains analog and digital circuits that operates from different supply levels. As a result, the internal power supply of the MCU has also to support a wide range of output voltage instead of designing separate power supply for each block which requires additional design and layout efforts. Moreover, depending on the performance requirements of the MCU or the mode of operation, the current consumption can vary very widely. It can be as high as 150-300 mA in active and high performance mode or it can be as low as 10-200 µA in sleep or idle mode. Consequently, the internal power supply of the MCU has to support a wide range of load currents. It is important to mention that since MCUs usually stay more than 50% of their time in sleep mode, the efficiency of their internal power has to be high not only in active mode (heavy load condition), but also in sleep mode (ultra-light load condition). Furthermore, each application puts different limitations and constrains on the passives (i.e. inductors or capacitors) used with the MCU. This includes different size and cost which exaggerate the constrains of the MCU's internal power supply which has to support a very wide range of passive components as well. Most importantly, since some low-noise MCUs usually contain noise sensitive IPs such as PLLs, Oscillators, and (open full item for complete abstract)

Committee: Ayman Fayed (Advisor); Patrick Roblin (Committee Member); Steven Bibyk (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, Case Western Reserve University, 2022, EECS - System and Control Engineering

In this thesis, a control methodology that merges the adaptive extremum seeking control (ESC) with the robust quantitative feedback theory (QFT) into one compact control scheme is proposed. It utilizes the properties inherent in both methods to develop a novel control law that guarantees the convergence of a systems performance function driven by a plant model to meet multiple robust control performance objectives simultaneously. The system structure is set up to have the output of a linear time-invariant (LTI) model with structured uncertainties as the operating variable for a nonlinear objective function with possibly time-varying parameters. A new bound called the extremum seeking bound (ESB) is introduced to the QFT design process. By utilizing the information derived from tuning the ESC scheme for the systems static function, an ESB can be constructed that accounts for the speed of convergence of the ESC scheme and then combined with the other classical QFT bounds to design a controller that is both adaptive and robust for the system structure presented. The control architecture for this design is presented and used to develop the conceptual basis and detailed mathematical justification for the control law. The robust extremum seeking control (RESC) design is validated on two engineering problems using MATLAB and Simulink. One is a nonlinear system with time-varying parameters driven by a dynamical model with uncertain parameters that is described at a high level of abstraction to demonstrate the general applicability of the design method. The second, a specific engineering problem, is the design of a generator torque controller of a 10 kilowatts (KW) wind turbine for maximum power point tracking (MPPT) in region 2 operation.

Committee: Mario Garcia-Sanz (Advisor); Christian Zorman (Committee Member); Evren Gurkan-Cavusoglu (Committee Member); Vira Changkong (Committee Member) Subjects: Applied Mathematics; Electrical Engineering; Engineering; Mechanical Engineering; Systems Design

Doctor of Philosophy, University of Akron, 2022, Electrical Engineering

This research proposes a direct voltage control approach for electric motors, including the single-stage converter topology and the control algorithms. The proposed motor drive system achieves smooth output voltage waveforms for phase excitations and utilizes them to extend the drive capability besides improving the torque ripple, noise, and vibration performance. Applicable to various motor types, the direct voltage control (DVC) is mainly investigated for driving switched reluctance motor (SRM) in the scope of this thesis. Different voltage regulation-based control algorithms are studied. Since the capability of shaping the phase voltage precisely allows control of any motor variables, this ability enables regulating the phase currents, flux linkages, and phase voltages to obtain superior performance. A finite element analysis (FEA) is performed to characterize the motor for building a dynamic simulation model for an SRM. The developed DVC and the conventional control are simulated using this machine model in comparison to each other. A new dual polarity power converter (DPC) is modeled, which can buck and boost the DC bus voltage and provides a variable voltage generation (VVG). The DPC can process power in both directions and provide a variable voltage in both positive and negative polarities at the motor windings. Following the DPC design process, power boards and gate driver boards are manufactured and populated as modular systems for individual motor phases. The developed converter model is customized and sized to construct a motor drive for the targeted operating conditions of the investigated SRM. It includes a control board to enable the 3-phase operation and a single DC bus as the power source for all three modular power converters. A resistive load setup is built to test the converter's performance. After verifying the DPC's performance for its designed load conditions and position-dependent dynamics, the motor tests are performed. The motor tests (open full item for complete abstract)

Committee: Yilmaz Sozer (Advisor); Patrick Wilber (Committee Member); Alper Buldum (Committee Member); Igor Tsukerman (Committee Member); J. Alexis De Abreu Garcia (Committee Member) Subjects: Aerospace Engineering; Alternative Energy; Electrical Engineering; Electromagnetics; Electromagnetism; Energy; Engineering; Technology

Master of Science, The Ohio State University, 2020, Electrical and Computer Engineering

Power is an important issue for a data center as the workloads grow rapidly, especially as machine learning services increase exponentially, and it is essential to control the performance of systems to save energy cost. For customers, inference latency of machine learning services is their key concern, but accelerating inference means to increase performance level of servers. Also, downgrading power consumption of a system leads to bad service. Namely, it is important to balance the interaction between power consumption and service level agreement simultaneously. This thesis proposes a system to control the power and inference latency of machine learning workloads, which are evaluated by MIMO control approach and proportional feedback control method. We also apply theoretical methodology to systematically design the control loop with assurance of stability and performance such as settling time and inference accuracy. Our results show that MIMO control gives better performance in terms of settling time and steady-state errors, and the error rate is less than 2% for latency control and less than 1% for power control, which are precise enough for most applications.

Committee: Xiaorui Wang (Advisor); Wladimiro Villarroel (Committee Member) Subjects: Computer Engineering; Electrical Engineering

Doctor of Philosophy (PhD), Wright State University, 2020, Electrical Engineering

The pulse-width modulated (PWM) dc-dc converters play a vital role in several industrial applications that include motor drives, electric vehicles, dc distribution systems, and consumer electronics. The switched-mode power converters step the input voltage up or down based on their typology and provide a regulated output voltage. The stability and regulation performance of a power converter can tremendously be improved via a suitable control design. However, due to the nonlinearity of the power converters and the presence of the line and load disturbances, the design of a robust and low-cost control circuit becomes a challenging task. The sliding-mode control of the dc-dc converters has been studied for decades because of its robustness, design simplicity, and suitability for variable structure systems. Despite the merits of the sliding-mode control method, the linear controllers are still dominant and attractive to the commercial applications since they require less design efforts and can be implemented using simple analogue circuits. This research aims to develop simplified sliding-mode control circuits for the classical PWM dc-dc converters in continuous-conduction mode (CCM). The control objectives are to maintain a constant switching frequency, enhance the transient response, provide wide operating range, and track the desired reference voltage under large disturbances. In order to design and test the control circuit, an accurate power converter model should be derived. Hence, large-signal non-ideal averaged models of dc-dc buck and boost converters in CCM are developed. The models are simulated in MATLAB/SIMULINK and compared with the corresponding circuits in SaberRD simulator for validation purpose. Next, PWM-based simplified sliding-mode voltage and current control schemes are designed for the dc-dc buck and boost converters in CCM, respectively. The design procedure and the analogue realization of the control equations are presented, where the control c (open full item for complete abstract)

Committee: Marian K. Kazimierczuk Ph.D. (Advisor); Raúl Ordóñez Ph.D. (Committee Member); Saiyu Ren Ph.D. (Committee Member); Yan Zhuang Ph.D. (Committee Member); Xiaodong Zhang Ph.D. (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, University of Akron, 2018, Electrical Engineering

This dissertation addresses advanced control methodologies for the five-phase permanent magnet assisted synchronous reluctance motor (F-PMa-SynRM) drive under various open phase fault conditions. F-PMa-SynRMs are principally reluctance-type machines which contain fewer magnets than permanent magnet machines. The major advantage of F-PMa-SynRMs is their inherent fault-tolerant capability, which makes them suitable for critical applications in the automotive and aerospace industries. However, under different open phase faults, F-PMa-SynRMs lose their primary advantages due to reduced average torque and higher torque ripple creating severe vibrations that may cause immediate system shutdown. Additionally, the parameter estimation becomes challenging due to the temperature variations and the presence of current harmonics. In these situations, it is essential to develop advanced fault-tolerant control (FTC) methods for F-PMa-SynRMs targeting the maximization of average torque, minimization of torque ripple, and accurate estimation of temperature. Also, during the FTC, a fault detector is necessary to implement any feedback control methods. An optimal phase advance control method is proposed to maximize the reluctance torque with a minimum phase current under different open phase fault coniii ditions. Then, an active torque ripple minimization (TRM) technique adopting three major steps is proposed as follows: (i) active current harmonic identification, (ii) percentile harmonic injection, and (iii) vector rotation of healthy phases. After that, an analytical method is proposed to estimate magnet temperature in the F-PMa-SynRM without needing any temperature sensors. Finally, this dissertation develops a simplified fault detection method based on five-phase symmetrical component (SC) theory. Extensive simulation using MATLAB and finite element method is done to validate the theoretical claims. For further validation, experiments have been conducted on a 2.9 kW F-PMa-S (open full item for complete abstract)

Committee: Seungdeog Choi (Advisor) Subjects: Engineering

Doctor of Philosophy, The Ohio State University, 2016, Electrical and Computer Engineering

Electric power grid is experiencing a major paradigm shift toward a more reliable, efficient, and environmentally friendly grid. The concept of microgrid is introduced to integrate distributed renewable generation in proximity to demands for both environmental and power-efficient promises. A microgrid can be disconnected, or "islanded", from the main grid and operates on its own, providing energy to remote areas or during faults of the main grid for better reliability. Islanded microgrids inherit several different properties from traditional power grids, including uncertain and limited generation, mixed R/X ratio lines, and lack of power inertia from synchronous generators. Those properties pose new challenges for the stable operation of islanded microgrids. The dissertation is dedicated to addressing the control challenges of islanded microgrids. The contribution is twofolds. First, we propose a polynomial time optimal power flow (OPF) solver which finds an optimal operating point for the inverters of the distributed energy resources. The proposed algorithm can account for the cost functions on the reactive generation that are common in microgrids. It also brings new understanding on the conjectures of exact semidefinite programming (SDP) convex relaxation on the OPF problem. Furthermore, we show that without the load over-satisfaction assumption usually seen in the literature, a near global optimum can be found for the OPF problem with arbitrary convex quadratic cost functions. The results are important to both microgrids and the classical OPF problem. Our second major contribution is developing a novel distributed controller that addresses the control challenges originated from limited generation, mixed R/X ratio lines, and lack of power inertia properties of islanded microgrids. The proposed controller can ensure proportional active and reactive power sharing and frequency synchronization while respecting the voltage constraints. Variances of the distributed (open full item for complete abstract)

Committee: Wei Zhang (Advisor); Kevin Passino (Committee Member); Andrea Serrani (Committee Member); Krishnaswamy Srinivasan (Other) Subjects: Electrical Engineering; Mechanical Engineering

Master of Science in Mechanical Engineering (MSME), Wright State University, 2016, Mechanical Engineering

One of the greatest challenges in additive manufacturing is fabricating titanium structures with consistent and desirable microstructure. To date, fully columnar deposits have been achieved through direct control of process variables. However, the introduction of external factors appears necessary to achieve fully equiaxed grain morphology using existing commercial processes. This work introduces and employs an analytic model to relate process variables to solidification thermal conditions and expected beta grain morphology at the surface of and at the deepest point in the melt pool. The latter is required in order to ensure the deposited microstructure is maintained even after the deposition of subsequent layers and, thus, the possibility of equiaxed microstructure throughout. By exploring the impact of process variables on thermal, morphological, and geometric trends at the deepest point in the melt pool, this work evaluates four commercial processes, estimates the range of process variables capable of producing fully equiaxed microstructure, and considers the expected size of the resultant equiaxed melt pool.

Committee: Nathan Klingbeil Ph.D. (Advisor); Joy Gockel Ph.D. (Committee Member); Raghavan Srinivasan Ph.D. (Committee Member) Subjects: Aerospace Materials; Engineering; Materials Science; Mechanical Engineering; Metallurgy; Morphology

Master of Science in Electrical Engineering (MSEE), Wright State University, 2015, Electrical Engineering

In this thesis, an average current-mode controller is analyzed for controlling power electronic converters. This controller consists of two loops. An inner loop, which senses and controls the inductor current and an outer loop, which is used to control the output voltage and provide reference voltage for the inner current loop. An average current-mode controller averages out high frequency harmonics it senses from the inductor current to provide a smooth DC component. This can be used as a control voltage for a pulse width modulator and produce switching pulses for the power electronic converters. An average current-mode controller can also be designed for a good bandwidth, which helps in accurate tracking of the sensed inductor current. For a better understanding of the operation of an average current-mode controller analytical equations are derived. Many transfer functions, which help analyze the properties of an open loop system, the controller transfer functions and a block diagram representing the converter along with current and voltage-control loops are presented. The block diagram and the transfer functions were used to derive the required controller parameters on MATLAB. The designed converter along with the controller is implemented on SABER circuit simulator. Waveforms representing the analytical equations along with the dynamic properties of the converter with the controller were plotted. The plotted SABER simulations were in agreement with the analytical equations. The designed controller was able to produce a controlled output voltage for step change in input voltage and load resistance, when simulated on SABER. Ripples could be observed in the control voltage of the controller, when designed for a good bandwidth. This was also represented by the derived analytical equations.

Committee: Marian K. Kazimierczuk Ph.D. (Advisor); Yan Zhuang Ph.D. (Committee Member); Lavern Alan Starman Ph.D. (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, University of Toledo, 2009, Electrical Engineering

Series hydraulic hybrid technology has the potential to significantly improve fuel economy and reduce emission. The series hydraulic hybrid is very different from electric and parallel hydraulic configuration and requires a unique power management control strategy to realize its optimal potential. In this dissertation, three approaches to achieve optimality are proposed and analyzed. These are rule-based, intelligent, and mixed power management control strategy. For evaluating the performance of control strategies, a forward-facing closed-loop simulation model based on physical features is first established in the MATLAB/SIMULINK environment. We then introduce a simple, valid and easily implementable rule-based power management control strategy. To derive the control signals, a PID-based multi-stage controller is presented. A thorough analysis on a class VI medium truck is elucidated. The simulation results demonstrate that a series hydraulic hybrid medium truck with the proposed rule-based power management control strategy results in fuel economy increases of 117% and 44% over the conventional baseline respectively over Federal Urban Driving Schedule (FUDS) and Federal Highway Driving Schedule (FHDS). Then, an intelligent power management control strategy incorporating artificial neural networks (ANNs) and dynamic programming (DP) algorithm applied to series hydraulic hybrid propulsion systems is presented. ANNs are used to forecast vehicle speed and DP is utilized to find the optimal control actions for gear shifting and dual power source splitting. A thorough analysis of effect on fuel economy with different prediction window size on the class VI medium truck over FUDS and FHDS is presented. Compared with conventional baseline, the simulation results demonstrate that series hydraulic hybrid medium truck with 20 seconds short-term prediction window enables fuel economy increase of 135% and 48% respectively over FUDS and FHDS. Although the intelligent power mana (open full item for complete abstract)

Committee: Roger King (Committee Chair); Walter Olson (Committee Co-Chair); Thomas Stuart (Committee Member); Richard Molyet (Committee Member); Gursel Serpen (Committee Member) Subjects: Electrical Engineering

Master of Science in Electrical Engineering, Cleveland State University, 2009, Fenn College of Engineering

In an interconnected power system, as a power load demand varies randomly, both area frequency and tie-line power interchange also vary. The objectives of load frequency control (LFC) are to minimize the transient deviations in theses variables (area frequency and tie-line power interchange) and to ensure their steady state errors to be zeros. When dealing with the LFC problem of power systems, unexpected external disturbances, parameter uncertainties and the model uncertainties of the power system pose big challenges for controller design. Active disturbance rejection control (ADRC), as an increasingly popular practical control technique, has the advantages of requiring little information from the plant model and being robust against disturbances and uncertainties. This thesis presents a solution to the LFC problem based on ADRC. The controller is constructed for a three-area power system with different turbine units including non-reheat, reheat and hydraulic units in different areas. The dynamic model of the power system and the controller design based on the model are elaborated in the thesis. Simulation results and frequency-domain analyses proved that ADRC controller is attractive to the LFC problem in its stability and robustness.

Committee: Lili Dong PhD (Committee Chair); Daniel Simon PhD (Committee Member); Zhiqiang Gao PhD (Committee Member); F. Eugenio Villaseca PhD (Committee Member) Subjects: Electrical Engineering

Master of Science, University of Akron, 2008, Electrical Engineering

The Z-source converter (ZSC) is an alternative power conversion topology that can both buck and boost the input voltage using passive components. It uses a unique LC impedance network for coupling the converter main circuit to the power source, which provides a way of boosting the input voltage, a condition that cannot be obtained in the traditional inverters. It also allows the use of the shoot-through switching state, which eliminates the need for dead-times that are used in the traditional inverters to avoid the risk of damaging the inverter circuit.Dynamic modeling of the ZSC from different perspectives has been studied in the literature. So far, based on these models, the peak dc-link voltage has been controlled using direct measurement, or through indirect control using measurement of the capacitor voltage. Direct measurement control requires a peak detection circuit due to the pulsating nature of the dc-link voltage. Indirect control using the capacitor voltage makes the peak dc-link voltage sensitive to line disturbances. In this research, the peak dc-link voltage is reconstructed using the measurements of the capacitor and input voltages. The ZSC is then controlled using two methods; namely the Voltage Mode (VM) and Current Programmed Mode (CPM). The two control laws for the ZSC with inductive loading are derived based on a small signal model of the converter. In CPM, since the order of the system is reduced by one, it is possible to achieve a dynamic performance similar to VM control with a simpler compensator. Performances of both VM and CPM controlled ZSC are verified by the simulation and experimental results for line and load disturbances. The simulation and experimental results for both steady state operation and disturbance rejection cases observed to be comparable. Line disturbance rejection of CPM controlled ZSC is found to be better than the VM control case. Both control methods showed satisfactory dynamics in case of the load disturbance rejectio (open full item for complete abstract)

Committee: Malik E. Elbuluk PhD (Advisor) Subjects: Electrical Engineering; Energy; Engineering; Technology

Master of Science, The Ohio State University, 2007, Graduate School

Committee: Not Provided (Other) Subjects:

Doctor of Philosophy, The Ohio State University, 2024, Electrical and Computer Engineering

Data centers often need to run lots of Machine Learning (ML) applications with stringent Service-Level Objective (SLO) requirements, such as inference latency. With bursty Machine Learning workloads, power capping is essential for high-density servers to safely oversubscribe the power infrastructure in a data center without upgrading expensive facilities. However, power capping is commonly accomplished by dynamically lowering the server processors' frequency levels, which can result in degraded ML application performance. Inference workload performance, such as recognition accuracy, must be optimized within a specific latency constraint, which demands high server power. Thus, the first challenge is providing SLO guarantees for inference requests on data center servers where power capping must be strictly enforced. In addition, data centers prefer to over-provision the number of servers used for inference processing and isolate them from other servers that run ML training, despite both using GPUs extensively, to minimize possible competition of computing resources. Those practices result in low GPU utilization and thus a high capital expense. Hence, the second challenge is to safely co-locate training and inference jobs on the same GPUs and servers with explicit SLO guarantees to improve GPU utilization and reduce capital expenses. In order to achieve the best inference accuracy under the desired latency and server power constraints, this dissertation proposes OptimML, a Multi-Input-Multi-Output (MIMO) control framework that jointly controls both inference latency and server power consumption, by flexibly adjusting the Machine Learning model size (and so its required computing resources) when server frequency needs to be lowered for power capping. Our results on a hardware testbed with widely adopted ML frameworks (including PyTorch, TensorFlow, and MXNet) show that OptimML achieves higher inference accuracy compared with several well-designed baselines while resp (open full item for complete abstract)

Committee: Xiaorui Wang (Advisor); Yang Wang (Committee Member); Marco Brocanelli (Committee Member) Subjects: Computer Engineering

Doctor of Philosophy, University of Akron, 0, Electrical Engineering

DC bus capacitor voltage unbalance in Neutral Point Clamped (NPC) multilevel inverters is an inherent challenge with a severe effect on five and higher-level inverters. Capacitor voltages unbalance depends on type of inverter loading and modulation techniques. Following this, modulation-based (soft) voltage balancing schemes manipulate modulations in accordance to load current conditions, while hardware-based (active) balancing, introduce external circuitry dedicated for voltage balancing. Current soft balancing schemes and modulations with unbalance mitigation for five-level NPC inverters require a high number of switchings, especially at high modulation index operations. Moreover, active balancing techniques come with high components count. This dissertation presents, three parts modulation and hybrid voltage balancing scheme, novel solutions for five-level NPC inverters to address limitations seen in earlier voltage balancing solutions. When implemented together, these solutions offer low switching operation and reduce voltage balancing circuitry. The three parts modulation combines three sub-modulations, which switch into different number of DC links within a switching period. These sub-modulations activate at different reference signal magnitude ranges. Three parts modulation reduces voltage unbalance in comparison to conventional modulation while offering soft balancing opportunities. Furthermore, the modulation offers operation with reduced switching when compared to state of the art carrier-based modulations suited for voltage balancing. To address voltage unbalance associated with the modulation, hybrid balancing scheme is preferred as it maintains inverter operation with low switching characteristics. These behaviors are analytically characterized, and verified using Matlab/SIMULINK simulations and experimental testing for both single-phase and three-phase inverters. The hybrid balancing uses active balancing and soft balancing techniques working togethe (open full item for complete abstract)

Committee: Malik Elbuluk (Advisor); Seungdeog Choi (Committee Co-Chair); Nao Mimoto (Committee Member); Kye-Shin Lee (Committee Member); Dane Quinn (Committee Member); Robert Veillette (Committee Member) Subjects: Electrical Engineering; Engineering