New Research Project HyProTwin
In an era of rapid technological innovation, the management and maintenance of transport infrastructure faces new challenges. The HyProTwin research project addresses exactly this, developing a method for the automated creation of digital twins of transport infrastructure for the process-oriented provision of heterogeneous existing data. In this blog post, we introduce our latest research project and explain what contribution plusmeta makes to the successful implementation of this ambitious endeavor.
Table of Contents
The Challenge: Confusing and Unstructured Existing Data for Transport Infrastructure
Transport infrastructure such as bridges, locks, weirs, and culverts are complex systems whose existing data comes in a wide variety of formats. Technical documentation, structural plans, maintenance logs, and sensor data are often scattered across different formats — sometimes as scanned documents, sometimes as unstructured datasets in various IT systems. This means that specialists frequently have to research, interpret, and compile this dispersed information manually. This process is not only time-consuming and error-prone, but also inefficient and therefore of little added value.
The Problem in Detail
1. Multitude of sources and formats: Existing data is often stored in different formats and at different locations, which makes consolidation difficult.
2. Considerable manual effort: Information is rarely available in a uniform and machine-readable form. Before these resources can be used, they frequently have to be prepared manually.
3. Lack of intelligent solutions: There is a lack of automated systems capable of intelligently linking and processing disparate data sources, such as those found in transport infrastructure data management.
These challenges underscore the need for innovative solutions to make technical existing data more efficiently accessible, thereby creating the foundation for well-founded decisions and the sustainable maintenance of valuable transport infrastructure.
Project Goal: Development of a Hybrid Digital Twin
The HyProTwin project aims to develop a solution that makes it possible to automatically create digital twins of transport infrastructure. Selected existing documents from structures on inland waterways — such as locks, weirs, and culverts — serve as sample data for the research project. The goal is to use digital twins to intelligently link current and historical data, making it more easily accessible and usable for future applications.
The HyProTwin Research Project Combines Two AI Approaches in an Innovative Way:
1. Symbolic AI: This method works with defined rules and logical structures that make it possible to map complex relationships and processes.
2. Sub-symbolic AI: This approach learns patterns from large amounts of data and identifies relationships that are not explicitly predefined.
This combined AI strategy enables a more detailed and comprehensive data analysis that goes far beyond the capabilities of conventional systems.
Technical writers and maintenance planners are also frequently confronted with obstacles regarding the quality and availability of the underlying information.
plusmeta's Contribution: Intelligent Documentation Solutions for HyProTwin
As an innovative software company, plusmeta plays a decisive role in the HyProTwin project. With our comprehensive expertise in the automated preparation of technical documents, we contribute our practical experience as a subcontractor to the Federal Waterways Engineering and Research Institute. We have already demonstrated in a joint pilot project with the Federal Waterways Engineering and Research Institute how the use of Artificial Intelligence can comprehensively optimize the assignment of metadata to existing documents in waterway infrastructure.
Integration of Metadata and AI Technologies
With its extensive experience, plusmeta supports the HyProTwin project in particular in the following areas:
1. Metadata pre-filtering: By implementing metadata to narrow down context (e.g., language, role, asset), relevant database entries can be precisely flagged. This enables more efficient data processing and retrieval.
2. Document localization: Our solution supports the precise search and localization of documents by integrating metadata in text form in the header information of data pieces or as separate data blocks.
3. Expanding the RAG architecture: By incorporating ontologies and knowledge graphs into the Retrieval Augmented Generation (RAG) architecture, plusmeta models knowledge comprehensively and enables an in-depth representation of knowledge.
Project Partners and the mFUND Funding Program
The HyProTwin project is carried out by a consortium of several partners:
CONTACT Software GmbH acts as the consortium coordinator of the project. As a leading provider of open standard software, CONTACT supports companies in their digital transformation.
The Federal Waterways Engineering and Research Institute (BAW) is a technical and scientific federal agency within the remit of the Federal Ministry for Digital and Transport (BMDV). As part of the HyProTwin project, plusmeta is involved as a subcontractor of the BAW.
The Institute for Numerical Methods and Computer Science in Civil Engineering (IIB) at TU Darmstadt serves as the scientific partner of the project.
Funding through mFUND
The project is funded by the mFUND innovation initiative of the Federal Ministry for Digital and Transport (BMDV). The mFUND funding program specifically promotes research and development projects dealing with digital, data-based applications for the mobility of the future. In addition to financial support, mFUND offers a wide range of event formats for networking among stakeholders from politics, industry, and research, as well as access to relevant data portals of the BMDV.
Implementation: Steps Toward Developing the Digital Twin
The implementation of the HyProTwin project is divided into several phases:
1. Analysis of information needs First, a detailed analysis is carried out to determine which information is required for the maintenance of transport infrastructure. This analysis forms the basis for developing the AI-powered search engine capable of automatically finding, structuring, and processing data from various sources in a process-oriented way.
2. Development of a conceptual information model Using a conceptual information model (ontology) that describes the meaning and relationships of the information, the digital twin is created. This model serves as a blueprint for the intelligent linking and accessibility of the various existing data.
3. Implementation and testing of the solution The developed solution is tested for practical suitability in collaboration with the Federal Waterways Engineering and Research Institute (BAW) on real waterway structures.
The developed solution is implemented and validated as a demonstration application.
Expected Results and Broader Impact
Through the successful implementation of the HyProTwin project, concrete results are targeted in the following areas:
- **Foundation for building digital twins:** Creation of hybrid AI methods that facilitate the creation and maintenance of digital twins of transport infrastructure.
- **Standardized ontology:** The use of ontologies contributes to the unification and standardization of knowledge representation in the field of transport infrastructure.
- **Transferability of concepts:** The developed concepts serve as a standard template for similar projects in other application areas, enabling broad application and scaling.
- **Efficient and economical operation of infrastructure:** Through digital networking and intelligent preparation of existing data, a long-term economical and efficient management of infrastructure is made possible.
This research project aims to sustainably improve the maintenance and management of transport infrastructure and lay the groundwork for modern, digital infrastructure management.
Conclusion
The HyProTwin project represents a significant step toward intelligent and efficient management of transport infrastructure. Through the development and combination of a standardized ontology, an innovative combination of various AI methods, and the innovative solutions of plusmeta, a promising foundation is being created for the development of digital twins. This concept facilitates access to important existing data, improves decision-making, and thus contributes to the sustainable maintenance of infrastructure.
We at plusmeta are proud to be part of this groundbreaking project and to contribute to the digitalization of our transport infrastructure. We are particularly pleased to have the support of our new working student Robert Hormann, who will assist us as a researcher on this project.
Research Project KREEN
plusmeta has its roots in science and research. That is why we continuously develop our products and work every day to make the latest technologies usable in practice. With the KREEN project, we already successfully researched AI applications for the research and extraction of the decision basis for the sustainable and resource-efficient operation of machines and systems.
