Optimal Modulation Techniques for Isolated Multi-Port Converters

Distributed energy resources (DERs), encompassing small-scale energy sources like solar panels and wind turbines, pose challenges and opportunities for power converters due to the diverse power and voltage levels coexisting in proximity. Ensuring an effective power routing between these ports, high efficiency, and high power density are the main key features for the required converter interfacing these ports. Notably, in some applications, like electric vehicles (EVs), galvanic isolation from the ac grid is required, further emphasizing the need for versatility and compactness. While the dual active bridge (DAB) shows promise in various applications, utilizing multiple two-port converters in DERs can increase overall size and compromise power density and efficiency. To tackle these challenges, the dissertation introduces isolated multi-port converters (IMPCs) as innovative solutions. IMPCs leverage shared magnetic concepts to enhance power density, efficiency, and galvanic isolation among ports, offering benefits such as simplified power routing and reduced components. The primary focus of the research is on studying and analyzing three-port, triple active bridge (TAB) and four-port, quadruple active bridge (QAB) converters. The objective is to optimize their performance, specifically emphasizing efficiency. The research begins with a literature review examining proposed optimization methods for TAB and QAB converters. Subsequently, a two-step optimization strategy is presented specifically for the TAB converter. Initially, a favorable modulation pattern for efficiency maximization is identified, followed by an analytical optimization of the total rms current, considering the previously identified optimal modulation patterns. Recognizing the complexity of mathematically modeling the optimization problem, an alternative modelfree approach is introduced, known as multi-dimensional ripple correlation control (MD-RCC). MD-RCC, leveraging extremum-seeking control (ESC) concepts, dynamically adapts to parameter changes during operation without requiring prior knowledge of the converter model. MD-RCC has undergone extensive and rigorous testing, employing various optimization cost functions, including total rms current, input dc current, fundamental rms current, and a specially defined cost function aimed at ensuring zero voltage switching (ZVS) operation with low total rms current. Notably, MD-RCC demonstrates its effectiveness in optimizing both TAB and QAB converters. The latest proposed optimization approach involves data-driven (DD) technique, utilizing artificial neural networks (ANN) to model the optimal operation of the TAB converter based on load conditions. The ANN is trained using a simulation model aligned with an experimental TAB prototype. Then the ANN performance is tested and validated on the experimental prototype. Each proposed optimization approach underwent rigorous testing using various simulation tools, including Matlab/Simulink and PLECS. Additionally, hardware-in-the-loop (HIL) real-time simulation was employed to validate the performance of the optimization methods. Experimental prototypes were designed and constructed in the laboratory, primarily focusing on TAB and QAB converters, each rated at 5 kW. These real-world prototypes played a crucial role in substantiating the effectiveness of the optimization strategies in practical operating conditions.