The heating and cooling supply systems of buildings are usually documented in schematics. These documents contain important information about the topology and the components of the systems that are useful for designers, building managers, and technical monitoring service providers along the complete building life cycle. However, these schematics are often only available in paper form, as CAD or pdf files and the information that they contain is almost not accessible digitally. This situation makes the data collection tedious and hampers the efficiency and cost-effectiveness of refurbishment or a technical monitoring projects.
In the "DiMASH" project, we are developing innovative methods to digitally capture, link and describe the heterogeneous information and relationships contained in HVAC schematics. To this, the developers are relying on innovative methods from image and text recognition and machine learning. The digitized schematics can subsequently be used to support a refurbishment project or a technical monitoring.
Digitized plant schematics, which document the heating and cooling supply of non-residential buildings, contain a lot of information about the individual components of the system and the relationships between different components. There is also information from on-site surveys. However, all this information, which corresponds to the “as-built” condition of the installed technical building systems, has so far had to be compiled and evaluated manually.
In the “DiMASH” research project, a team of researchers and planners is developing a digital process to record and link the heterogeneous information and correlations contained in a plant schematic. The developers are relying on innovative methods from image and text recognition and machine learning.