Semantic digital twin of a university building
Machine learning and consumption monitoring Active
This project builds a semantic digital twin of a real UQTR building. Live BACnet point values stream through MQTT into an RDF triple store, where they are described with four complementary building-domain ontologies: Brick, ASHRAE 223P, RealEstateCore and Haystack.
The knowledge graph is the central artefact: it is validated with SHACL, queryable in SPARQL and exposed as a natural-language question-answering service backed by a large language model (GraphRAG). Design priorities are, in order: semantic correctness, interoperability, operational simplicity and cost.
Partners
- UQTR