Session

Artificial intelligence (AI) and machine learning (ML) have the potential to enable significant reductions in energy use by finding hidden connections in building data and catching operational issues that might otherwise go undetected. Yet, many buildings do not have the near-perfect data that is required to deploy the ML algorithms found in the literature. Real-world data quality and availability limitations may constrain the available options when considering ML-based use cases.

This session will define AI and ML, discuss common applications in the literature, and explore a case study where ML is being applied to real-world buildings data. The discussion will include examples of common ML use cases, such as baseline energy prediction and fault detection. It also will explore the challenges of applying ML algorithms developed using high-quality datasets to problems using messy and incomplete real-world datasets. At the end of the session, participants will gain an understanding of the opportunities and challenges associated with applying ML to real-world buildings data.