| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Advanced Materials for Energy Conversion and Storage 2025
|
| Presentation Title |
Exploring Large Parameter Spaces of Energy Storage using Domain-Knowledge-Informed Machine Learning |
| Author(s) |
Maher Alghalayini, Daniel Collins-Wildman, Kenny Higga, Vincent Battaglia, Stephen Harris, Marcus Noack |
| On-Site Speaker (Planned) |
Maher Alghalayini |
| Abstract Scope |
Long-duration energy storage systems are used to mitigate the increase in global warming by integrating renewable energy sources into the grid. These systems operate over extended durations, and their performance depends on many parameters. Optimizing these systems requires efficiently exploring their parameter spaces and assessing durability under real-life cycling. Traditionally, this exploration process has been inefficient because laboratory cycling uses identical cycles, which are not representative of real-world use, resulting in inaccurate durability prediction. Here, we present an innovative method that harnesses statistical and machine learning techniques, specifically Gaussian process regression enhanced with domain knowledge and expected information gain from future experiments, to streamline the exploration of parameter spaces under real-world cycling. Results show that our approach efficiently explores the parameter space and approximates durability with minimum testing. In short, this work holds promise to expedite the development of energy storage, facilitating renewable energy integration and a more sustainable future. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Energy Conversion and Storage, Machine Learning, |