I-Corps: Machine Learning Enhanced Automated Circuit Configuration and Evaluation of Power Converters

The broader impact/commercial potential of this I-Corps project is the development of power electronics to automate complex circuit design. While significant progress has been made in advancing modeling simulation and verifying electrical power converters, the process of designing such devices remains inefficient in terms of time and cost. The state-of-the-art circuit design of power converters relies heavily on human experts to select the optimal topology and search for design parameters with human experience and intuitions. This process can be very time-consuming, inefficient, and labor-intensive. The proposed software may help power electronics engineers consider a wide range of novel concepts more rapidly and cost-effectively before selecting an engineering-optimal architecture for high-fidelity design and evaluation. This I-Corps project is based on the development of technology that integrates recent breakthroughs in machine learning, power electronics, data analytics, simulation software, and optimization to automate the circuit design of electrical power converters. The technology may also facilitate the integration of the proposed software tools into existing power-converter design workflows. The technology seeks to automatically generate and evaluate power converter designs with physics-based Reduced Order Models: automatically evolving architecture concepts toward the optimal system configurations and automatically generating, evaluating and optimizing architectures within acceptable performance uncertainties while satisfying the desired outputs. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.