Towards standard data organization for data-driven building applications in cyber-physic systems

Abstract

With the advance of data and AI technology, the built-in environment is increasingly digitalised and computerised. A substantial amount of data from the heterogeneous building systems, devices and sensors can be gathered, stored, analysed and utilised. Emerging building applications rely on data-driven machine learning (ML) algo­rithms. However, developing these applications is still challenging—developers need to have the knowledge about what ML models and building data to use as well as how these data are organized in each individual building, despite the existing solu­tions (e.g. standard building metadata schema, Brick) to standardizing data and resource management in buildings. There is a gap between building applications de­velopment and the data organization in the building systems, which heavily impedes the portability and extensibility of building applications. This dissertation posits that standardising building data organisation presents a po­tential solution to overcoming existing barriers of building application development. By constructing building metadata models following established standard schemes, the development of portable building applications can be facilitated. Extending the existing standard for building metadata schema enhances the extensibility of building applications and enables involvement of sectors from other domains, such as energy storage systems and electric vehicles. To this end, this thesis first proposes Cloze - a building metadata model generation system that can convert the metadata of a building in an ad hoc convention into a building metadata model that follows a standard building data schema. To access the building data for portable building applications, this thesis proposes EnergonQL, that is a query language that extracts data for building applications, agnostic to building specifics such as its subsystems and structure. A data acquisition system for build­ings, Energon, has been implemented using EnergonQL. Energon enables developers to execute EnergonQL queries with a standard building metadata model that has been generated by Cloze. Lastly, the thesis proposes BESOI, an ontology integration system that extends existing ontology of built-in environments with ontology from other domains, such as energy storage system.