Nigel Groome Studentship: AI-Enhanced Battery State of Health Estimation Using Ring Probabilistic Logic Neural Networks

Accurate estimation of battery State of Health (SOH) is vital for safety, performance, and longevity in electric vehicles and energy storage systems. Current models struggle to balance accuracy, generalisability, and computational efficiency across diverse operating conditions.

This project proposes a novel AI-based framework, the Ring Probabilistic Logic Neural Network (RPLNN), which fuses probabilistic logic and neural computation to enhance SOH prediction robustness and interpretability. The project will develop, train, and experimentally validate the RPLNN model using lithium-ion cell data supplied by Jaguar Land Rover (JLR).

Unlike conventional deep models that learn opaque mappings, the RPLNN constrains information flow through a ring-based structure governed by probabilistic logic rules. This approach improves interpretability, data efficiency, and resistance to data drift, addressing key limitations in current AI-based SOH methods.

The studentship requires you to undertake the equivalent of up to 6 hrs of teaching per week on average, during semester time, and to include preparation and marking (but no more than 20 hrs per week), and to participate in a teaching skills course without further remuneration.