AI agent-based indoor environmental informatics: Concept, methodology, and case study

Analyzing and improving indoor environments requires continuous intervention from human experts, which is challenging in practice. To address this limitation, an ontology-based AI agent system can be employed. This study proposes a conceptual model of AI agent-based indoor environmental informatics (IEI), develops an indoor environmental ontology by extending Brick schema, and demonstrates its application in real indoor environment. AI agent-based IEI is an approach that builds an integrated information system to capture relationships among indoor environments, occupants, and building systems so that utilizes AI agent for continuous indoor environment management. The AI agent leverages indoor environmental ontology, intrusive data, and indoor environmental toolkit to perform holistic analysis of indoor thermal environments and provides strategies for enhancing thermal comfort during the operational phase. The proposed concept was implemented in the Dynamo environment and applied to an office space. For the collection of real indoor environmental data, intrusive measurement was conducted over five days, and an indoor environmental ontology for the target space was developed. Indoor environmental toolkit used for the system included spatial coordinate extractor (SCE) for extracting spatial element coordinates, ontology file generator (OFG) for creating ontology files, and predicted mean vote (PMV) model for calculating PMV. The AI agent identified a PMV variation of 0.77, a discomfort rate of 28.2 %, and the disparity between physical sensor data and occupants’ subjective thermal comfort. Furthermore, the AI agent suggested practical strategies for improvement, including determining window status based on outdoor temperature, adjusting air conditioner operation, and modifying occupant seating arrangements.