Machine learning for guiding high-temperature PEM fuel cells with greater power density

Renewable energy and energy efficiency are crucial for achieving global sustainability goals. In this context, there is need for the development of new materials that realize high-performing and low-cost power sources. At the same time, advances in computational power, simulation, and Machine Learning enable researchers to explore large amounts of data, providing inspiration and tools for the design of new systems. In this work, we combined experiments with modeling and data analysis tools to build a framework for the study and development of high-temperature polymer electrolyte membrane fuel cells (HT-PEMFCs). The framework used Machine Learning tools (e.g., support vector regression, dimension reduction, and clustering) that seamlessly linked materials characteristics with fuel cell performance. This allowed for the accelerated discovery of material properties and fuel cell operating parameters that achieve greater power density while co-currently addressing costs.