News
I presented our poster, “Lightweight Physics-Informed Reservoir Computing for Battery Health Prediction,” at the 2026 American Control Conference (ACC) in New Orleans, Louisiana, USA.
Predicting the State of Health (SOH) of lithium-ion batteries is safety-critical to prevent thermal runaway and extend lifespans in electric vehicles. Current models often force a compromise between computational speed and physical accuracy. To bridge this gap, we introduced the PIRC (Physics-Informed Reservoir Computing) framework.
By combining:
- Explicit polynomial NVAR features with recurrent reservoir states
- Closed-form ridge regression solver to avoid costly backpropagation
PIRC enables energy-efficient, accurate SOH monitoring directly on edge hardware. This approach achieves better prediction accuracy while remaining computationally lightweight and suitable for real-time deployment in battery management systems.
A massive thank you to Prof. Yanwen Xu and Prof. Wenbin Wan for their guidance and support.
It was also amazing to see some incredible robotics displays on the conference floor!