Closing the Expertise Gap in Residential Building Energy Retrofits: A Domain-Specific LLM for Informed Decision-Making
#Large Language Model #Energy Retrofits #Residential Buildings #CO2 Reduction #Techno-economic Analysis #Low-Rank Adaptation #Decarbonization
📌 Key Takeaways
- Researchers developed a domain-specific LLM to address the expertise gap in residential energy retrofit decisions
- The model uses homeowner-accessible descriptions to provide optimal retrofit recommendations
- Fine-tuning on 536,416 U.S. residential building prototypes achieved 98.9% accuracy in CO2 reduction optimization
- The model maintains performance with incomplete input data, supporting real-world application
📖 Full Retelling
Researchers led by Lei Shu have developed a domain-specific large language model to help homeowners make informed decisions about residential building energy retrofits, addressing the expertise gap that prevents many from understanding complex energy assessments. The model, fine-tuned on physics-based energy simulations and techno-economic calculations from 536,416 U.S. residential building prototypes across nine major retrofit categories, uses Low-Rank Adaptation to map dwelling characteristics to optimal retrofit selections. Published on February 19, 2026, the study demonstrates that the model identifies optimal retrofits for CO2 reduction within its top three recommendations in 98.9% of cases and shortest discounted payback period in 93.3% of cases.
The research addresses a significant challenge in residential energy efficiency: homeowners typically lack the technical expertise required to evaluate which retrofits would provide the best energy savings and return on investment. The developed LLM bridges this gap by accepting simple, homeowner-friendly descriptions of basic dwelling characteristics and providing tailored recommendations. Unlike generic AI models, this domain-specific solution incorporates physics-based energy simulations and economic calculations, ensuring that its suggestions are both technically sound and financially viable.
The model's performance was rigorously evaluated against physics-grounded baselines, showing remarkable accuracy in identifying optimal retrofits. Fine-tuning the model resulted in an order-of-magnitude reduction in CO2 prediction error and multi-fold reductions for energy use and retrofit cost predictions. Importantly, the model maintains high performance even when provided with incomplete information about a dwelling, making it practical for real-world applications where homeowners may not have access to comprehensive building data. This capability supports informed residential decarbonization decisions at scale, potentially accelerating the transition to more energy-efficient housing stock.
🏷️ Themes
Artificial Intelligence, Energy Efficiency, Home Renovation
📚 Related People & Topics
Climate change mitigation
Actions to reduce net greenhouse gas emissions to limit climate change
Climate change mitigation, also called climate change decarbonisation, is an action to limit the greenhouse gases in the atmosphere that cause climate change. Climate change mitigation actions include conserving energy and replacing fossil fuels with clean energy sources. Secondary mitigation strate...
Large language model
Type of machine learning model
A large language model (LLM) is a language model trained with self-supervised machine learning on a vast amount of text, designed for natural language processing tasks, especially language generation. The largest and most capable LLMs are generative pre-trained transformers (GPTs) that provide the c...
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Computers and Society arXiv:2602.20181 [Submitted on 19 Feb 2026] Title: Closing the Expertise Gap in Residential Building Energy Retrofits: A Domain-Specific LLM for Informed Decision-Making Authors: Lei Shu , Armin Yeganeh , Sinem Mollaoglu , Jiayu Zhou , Dong Zhao View a PDF of the paper titled Closing the Expertise Gap in Residential Building Energy Retrofits: A Domain-Specific LLM for Informed Decision-Making, by Lei Shu and 4 other authors View PDF Abstract: Residential energy retrofit decision-making is constrained by an expertise gap, as homeowners lack the technical literacy required for energy assessments. To address this challenge, this study develops a domain-specific large language model that provides optimal retrofit recommendations using homeowner-accessible descriptions of basic dwelling characteristics. The model is fine-tuned on physics-based energy simulations and techno-economic calculations derived from 536,416 U.S. residential building prototypes across nine major retrofit categories. Using Low-Rank Adaptation , the LLM maps dwelling characteristics to optimal retrofit selections and associated performance outcomes. Evaluation against physics-grounded baselines shows that the model identifies the optimal retrofit for CO2 reduction within its top three recommendations in 98.9% of cases and the shortest discounted payback period in 93.3% of cases. Fine-tuning yields an order-of-magnitude reduction in CO2 prediction error and multi-fold reductions for energy use and retrofit cost. The model maintains performance under incomplete input conditions, supporting informed residential decarbonization decisions. Comments: A preprint version is available via SSRN (Elsevier Preprint Service) Subjects: Computers and Society (cs.CY) ; Artificial Intelligence (cs.AI) Cite as: arXiv:2602.20181 [cs.CY] (or arXiv:2602.20181v1 [cs.CY] for this version) https://doi.org/10.48550/arXiv.2602.20181 Focus to learn more arXiv-issued DOI via DataCit...
Read full article at source