AI technologies have proven effective in enhancing building operations, conserving energy, and reducing carbon emissions. While existing ML model advancements typically focus solely on snapshot accuracy, practical application necessitates consideration of various aspects such as maintainability, reliability, and efficiency. Unfortunately, there is currently no comprehensive solution to understand the behavior of ML models, resulting in a lack of active utilization for the vast number of developed models. This project aims to develop a methodology and platform called BaiTest for evaluating AI models in buildings. BaiTest facilitates comparison and selection of suitable ML models from a vast array of options. Preliminary results demonstrate that choosing a model with high maintainability or reliability can yield energy savings of 3%-10% more than a model with the highest snapshot accuracy.
|Effective start/end date||1/05/23 → 30/05/25|
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.