Real-Time Active Thermal Management and Degradation Forecasting of Power Electronics Building Blocks In All-Electric Ships

With advancements in semiconductor technology, power electronic devices play a critical role in modern power systems, enabling efficient energy conversion in applications such as renewable energy integration, electric vehicle drives, industrial automation, and Navy ship power systems. However, the increasing use of power converters presents challenges such as harmonic reduction, cost optimization, reliability improvement, and thermal management, especially in weight- and space-constrained environments like wind turbines and shipboard systems. High-frequency switching, while reducing passive filter size, increases thermal stress on semiconductor devices, impacting efficiency and longevity. Active thermal control, which regulates junction temperature through power loss management, and degradation forecasting, which aids future maintenance decision-making, are key to enhancing semiconductor reliability and operational efficiency. This dissertation presents a comprehensive study of thermal control and management strategies in power electronics, with a focus on their application in All-Electric Ship (AES) power systems. It begins by reviewing existing electro-thermal management techniques and identifying their limitations in mission-critical, space-constrained environments. The dissertation then introduces a novel Finite Control Set Model Predictive Control (FCS-MPC) strategy that simultaneously regulates thermal behavior and controls the electrical performance of Modular Multilevel Converter (MMC)-based Power Electronic Building Blocks (PEBBs). In addition, a data-driven framework utilizing Deep Neural Networks (DNNs) is developed to forecast PEBB degradation, supporting more accurate and informed electro-thermal management decisions. By integrating real-time thermal regulation, enhanced electrical control, and predictive maintenance planning, this work aims to improve both the reliability and operational efficiency of AES power systems.