Solid State Transformer with Predictive Digital Twin

The modern electric power grid is facing increasing stress due to a fundamental shift in supply and demand technologies. These increased stresses require grid modernization to shift from analog systems to systems with increased data streams and digital controls. One such technology that can be utilized to accomplish these goals is the solid-state transformer which provides isolation between low and highvoltage connections by electronics-based power conversion, with a high frequency transformer. However, one concern that may arise with using solid-state transformer over legacy systems is increased reliability challenges in which factors such as use case and operational conditions can lead to unpredictable behavior. The overall objective of this program is to design, develop, and demonstrate a predictive digital twin for a solid-state transformer. Around the solid-state transformer hardware architecture and simulation models, a dynamic data-driven framework will be developed to provide real-time prognostic health monitoring and dynamic decision-making suggestions to the electrical grid controller and maintenance crew. These estimations enable maintenance strategies that can focus repairs and replacements of components nearing the end of their total life cycle, ultimately reducing the cost passed on to the public. This technology will improve the resiliency of the grid, reducing total power outages which cost the country 18 to 33 billion dollars annually. During the Phase I project, a physics-based simulation model of an existing solid-state transformer was developed with an algorithm to simulate the aging effects of silicon-carbide components. This model was used to develop a machine learning model that accurately predicts the remaining useful life of the system. The Phase II project will create an accelerated aging test rig and experimentally generate aging datasets. This database will then be used to refine and update the predictive digital-twin and add additional capabilities for monitoring the remaining useful life of the system. Additionally, the predictive-digital twin created during the Phase I of this program will be integrated into hardware and experimentally validated using a hardware-in-loop system. Other commercial applications for this technology include power electronics used in electric vehicle and electric aircraft powertrains, battery chargers, and solar inverters.