Building Decarbonization Planning Applying Machine-learning Techniques to Building Data

In response to the escalating demand for clean tech and energy-efficient structures, our project focuses on revolutionizing building data processing for swift decarbonization planning. We tackle the challenge by parsing diverse data from mixed media documents and fine-tuning pre-trained large language models (LLMs) for optimal data query accuracy. Our approach integrates text parsing, tokenization, and Computer Vision models to handle various files, encompassing text, graphical charts, and images. The project’s key outcome is a versatile data parser seamlessly combining text and image processing, generating datasets for LLMs. This innovation, along with LLM evaluation and parameter tuning, will be integrated into a modular framework for expedited and efficient building energy efficiency analysis. Ultimately, our solution aims to significantly hasten the delivery of impactful decarbonization plans, addressing a critical industry need.

Faculty Supervisor:

Michal Aibin

Student:

Partner:

SISA Energy

Discipline:

Computer science

Sector:

Professional, scientific and technical services

University:

British Columbia Institute of Technology

Program: