Metadata schema for virtual sensors in digital twin-enabled building systems using Brick schema

초록

Building digital twins (BDTs) require semantic interoperability to integrate physical and virtual components for effective building operation and management (O&M), yet the absence of a standardized metadata schema for representing virtual sensors creates ambiguity and constrains automation. This research addresses this problem by proposing a formalized ontology framework that extends the Brick schema to explicitly model virtual sensors. The aim of the study is to define a structured representation of virtual sensors that supports their semantic identification, relational mapping, and machine-interpretable integration into BDTs. The methodology involved introducing a new class for virtual sensors within Brick schema, defining associated properties and semantic relationships, and validating the extension through two case studies on a heating, ventilation, and air conditioning (HVAC) system equipped with an observation virtual sensor and a backup virtual sensor. Generative pre-trained transformer (GPT)-based artificial intelligence (AI) agents were employed to generate ontology instances and retrieve system metadata, with SPARQL queries used to analyze semantic and relational data. The findings reveal that the proposed Brick schema succeeded representing 16.7% of system instances linked to virtual sensors, while the original schema could not identify 21 relationships centered around virtual sensors, hindering complete semantic representation. The research demonstrates that the proposed Brick schema enhances interoperability, reasoning capability, and automation potential in digital twin–based building O&M. The novelty of this study lies in formally integrating virtual sensors into the Brick ontology, contributing a standardized and machine-actionable metadata framework for next-generation intelligent building systems.

키워드