Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Match Made in Semantics: Physics-infused Digital Twins for Smart Building Automation

Ganesh Ramanathan, Simon Mayer

TL;DR

The paper tackles the costly manual process of selecting appropriate reusable control programs for building automation by introducing PhyDiT, a bridging ontology that links control program descriptions, system designs, and underlying physical processes. It leverages physics-infused Web of Things TDs to embed process knowledge in decentralized, self-describing digital twins, enabling rule-based matching between technical systems and control programs. Through a real-world evaluation of 34 air-handling units, the approach achieves over 90% correct automatic matches, reducing manual effort substantially and illustrating the value of integrating physical-process knowledge into CP-TS reasoning. The work demonstrates that physics-aware semantic integration across CPD, SDD, and PPD, bridged by PhyDiT and embedded in TDs, can significantly advance autonomous building automation and support broader domains with similar multi-vendor ontologies.

Abstract

Buildings contain electro-mechanical systems that ensure the occupants' comfort, health, and safety. The functioning of these systems is automated through control programs, which are often available as reusable artifacts in a software library. However, matching these reusable control programs to the installed technical systems requires manual effort and adds engineering cost. In this article, we show that such matching can be accomplished fully automatically through logical rules and based on the creation of semantic relationships between descriptions of \emph{physical processes} and descriptions of technical systems and control programs. For this purpose, we propose a high-level bridging ontology that enables the desired rule-based matching and equips digital twins of the technical systems with the required knowledge about the underlying physical processes in a self-contained manner. We evaluated our approach in a real-life building automation project with a total of 34 deployed air handling units. Our data show that rules based on our bridging ontology enabled the system to infer the suitable choice of control programs automatically in more than 90\% of the cases while avoiding almost an hour of manual work for each such match.

A Match Made in Semantics: Physics-infused Digital Twins for Smart Building Automation

TL;DR

The paper tackles the costly manual process of selecting appropriate reusable control programs for building automation by introducing PhyDiT, a bridging ontology that links control program descriptions, system designs, and underlying physical processes. It leverages physics-infused Web of Things TDs to embed process knowledge in decentralized, self-describing digital twins, enabling rule-based matching between technical systems and control programs. Through a real-world evaluation of 34 air-handling units, the approach achieves over 90% correct automatic matches, reducing manual effort substantially and illustrating the value of integrating physical-process knowledge into CP-TS reasoning. The work demonstrates that physics-aware semantic integration across CPD, SDD, and PPD, bridged by PhyDiT and embedded in TDs, can significantly advance autonomous building automation and support broader domains with similar multi-vendor ontologies.

Abstract

Buildings contain electro-mechanical systems that ensure the occupants' comfort, health, and safety. The functioning of these systems is automated through control programs, which are often available as reusable artifacts in a software library. However, matching these reusable control programs to the installed technical systems requires manual effort and adds engineering cost. In this article, we show that such matching can be accomplished fully automatically through logical rules and based on the creation of semantic relationships between descriptions of \emph{physical processes} and descriptions of technical systems and control programs. For this purpose, we propose a high-level bridging ontology that enables the desired rule-based matching and equips digital twins of the technical systems with the required knowledge about the underlying physical processes in a self-contained manner. We evaluated our approach in a real-life building automation project with a total of 34 deployed air handling units. Our data show that rules based on our bridging ontology enabled the system to infer the suitable choice of control programs automatically in more than 90\% of the cases while avoiding almost an hour of manual work for each such match.
Paper Structure (16 sections, 5 equations, 2 figures, 4 tables)

This paper contains 16 sections, 5 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Technical Systems
  4. Control Programs
  5. Physical Processes and Mechanisms
  6. Web of Things and its Relevance to Automation Systems
  7. Approach
  8. The Need For Knowledge About Physical Processes
  9. The PyhDiT Ontology
  10. The Required High-level Abstractions
  11. Linking System Design Description to Physical Process Model
  12. Linking Descriptions of Control Program and System Design
  13. Synthesizing TDs
  14. Matching TDs to CPDs
  15. Evaluation
  16. ...and 1 more sections

Figures (2)

  • Figure 1: Overview of how the integrated knowledge in the two design workflows are used for automated matching.
  • Figure 2: Overview of the PhyDiT ontology. The bold lines and borders indicate the bridging concepts we have introduced. The dotted lines indicate relationships that are inferred (e.g., atComponent is inferred if the variable is associated to mechanism that is managed by the component.)