DiffBatt has been accepted for the Foundation Models for Science Workshop at NeurIPS 2024!
Our paper, DiffBatt: A Diffusion Model for Battery Degradation Prediction and Synthesis, has been accepted for the Foundation Models for Science Workshop at NeurIPS 2024!
In our research, we tackled the challenging problem of predicting how batteries degrade over time, a critical issue for advancing green technology and sustainable energy. Our solution, DiffBatt, utilizes advanced techniques from GenAI to accurately model and predict battery health.
Key Points:
- Probabilistic & Generative: DiffBatt captures the uncertainties in battery aging and simulates these changes over time effectively.
- Top Performance: It outperforms other models by predicting battery life with great accuracy across several datasets.
- Scalable & Expressive: Based on diffusion models, DiffBatt uses classifier-free guidance to generate robust, high-quality degradation curves.
We believe DiffBatt is a crucial step toward establishing a foundational model for battery degradation. By training on diverse datasets, it offers strong generalizability and robustness, paving the way for more sustainable tech solutions.
Excited to dive deeper? Discover how DiffBatt leverages GenAI to enhance battery technology in our paper and GitHub repo!