Neuro-Symbolic AI for Causal Reasoning with Large Language Models for Energy Storage Systems

This research project aims to advance the development of reliable, interpretable, and causally grounded Artificial Intelligence (AI) methods by integrating neuro-symbolic reasoning into Large Language Models (LLMs), with a specific application to energy storage systems such as lithium-ion battery systems. While modern LLMs demonstrate impressive linguistic capabilities, they remain limited in their ability to reason consistently over structured scientific knowledge and physical constraints, which is critical in safety-critical engineering domains. The project addresses this limitation by embedding symbolic knowledge, such as ontologies, causal rules, and physical laws, directly into transformer-based architectures to support transparent and scientifically valid reasoning. Through a combination of ontology construction, causal modeling, and hybrid neural-symbolic inference, the research will develop and evaluate a framework capable of reducing hallucinations, improving reasoning consistency, and aligning AI-generated explanations with established battery science. Conducted as part of a Master’s thesis and in close collaboration between the home institution and the Institute of Applied Artificial Intelligence at TELUQ, the project contributes to emerging research on trustworthy AI while supporting Canada’s innovation priorities in explainable artificial intelligence, energy storage technologies, and sustainable engineering systems.

Faculty Supervisor:

Belkacem Chikhaoui

Student:

Partner:

Lebanese American University

Discipline:

Engineering

Sector:

Artificial Intelligence; Green/Alternative Energy; Sustainability and the Environment

University:

Université TÉLUQ

Program: