Summary

Artificial intelligence (Al) is being used in many ways across the building industry. This presentation dives into real world applications of Al in Controls including experiences, successes, and lessons learned. Technologies discussed are practical applications of machine learning, including some that connect directly to contemporary generative AI models. The importance and basics of semantic data modeling and metadata management within building automation systems (BAS) to scale AI are also demonstrated at a high level.

This presentation also dives into different applications of machine learning and Al with real-life examples of camera-free occupant estimation via sensor fusion, cluster-based alarm management, BAS chat systems using large language models, and metaheuristic predictive control. The goal is to simplify the Al that is out there in the wild and show that it can be an economical path toward significant energy savings and operational improvements if executed effectively. Requirements and specifications for AI are also touched upon.

Speaker Bio

Jayson Bursill is the Data Scientist for Delta Controls where he leads all data science and related research activities. Jayson received his PhD in Mechanical Engineering from Carleton University for developing and implementing model-based predictive HVAC controls and evaluating performance in a field test. He is also an Adjunct Research Professor at Carleton University. He has publications on multiple building-related topics in several academic journals and conference proceedings. In addition to his experience and expertise, Dr. Bursill is also involved in several ASHRAE technical activities related to controls including TC 1.4, GPC 36, SPC 223P, SPC 231P, and MTG.GenAI.