AI-driven optimization framework with domain-specific Large Language Models for renewable energy and hydrogen deployment

This thesis introduces a novel framework for achieving zero-net emissions through Hybrid Renewable Energy Systems (HRES). It presents RE-LLaMA, a domain-specific Large Language Model trained on renewable and hydrogen energy data. The framework integrates RE-LLaMA with MG-OPT, a multi-objective optimization system that employs Mixed-integer linear Programming and Pareto-based optimization techniques. The framework uniquely combines photovoltaic, wind, and fuel cell sources with hydrogen production and storage capabilities while simultaneously optimizing four key objectives: minimizing system costs and emissions while maximizing renewable energy utilization and surplus profit through transactive energy mechanisms. Through comprehensive validation across three scenarios (30-50 kW systems), including EV and hydrogen fuel cell vehicle integration, the framework demonstrates remarkable environmental and economic benefits, achieving up to 95% emissions reduction and generating daily profits of $31-$45 through surplus electricity trading. The RE-LLaMA system's effectiveness is validated through LLM-as-Judge and human evaluation methodologies, with RE-LLaMA significantly outperforming the base model in renewable and hydrogen energy tasks, while MG-OPT's sophisticated algorithms enable precise resource allocation in grid-connected operation, ultimately advancing the field by bridging the gap between theoretical multi-objective optimization and practical implementation requirements for sustainable energy systems.

Name thesis:degree_name

Master of Applied Science (MASc)

Discipline thesis:degree_discipline

Software Engineering

Grantor

University of Ontario Institute of Technology

Year dc:date.issued

2025

Author dc:creator

• Hemied, Omar S.

Advisor dc:contributor.advisor

• Gaber, Hossam

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/10155/1931

OAI identifier oai:identifier

oai:ontariotechu.scholaris.ca:10155/1931