Filtern
Dokumenttyp
- Vortrag (10)
- Zeitschriftenartikel (2)
- Beitrag zu einem Tagungsband (1)
- Posterpräsentation (1)
- Forschungsdatensatz (1)
Schlagworte
- Research data management (15) (entfernen)
Organisationseinheit der BAM
Structural health monitoring (SHM) intends to improve the management of engineering structures. The number of successful SHM projects – especially SHM research projects – is ever growing, yielding added value and more scientific insight into the management of infrastructure asset. With the advent of the data age, the value of accessible data becomes increasingly evident. In SHM, many new data-centric methods are currently being developed at a high pace. A consequent application of research data management (RDM) concepts in SHM projects enables a systematic management of raw and processed data, and thus facilitates the development and application of artificial intelligence (AI) and machine learning (ML) methods to the SHM data. In this contribution, a case study based on an institutional RDM framework is presented. Data and metadata from monitoring the structural health of the Maintalbrücke Gemünden for a period of 16 months are managed with the RDM system BAM Data Store, which makes use of the openBIS data management software. An ML procedure is used to classify the data. Feature engineering, feature training and resulting data are performed and modelled in the RDM system.
Structural health monitoring (SHM) intends to improve the management of engineering structures. The number of successful SHM projects – especially SHM research projects – is ever growing, yielding added value and more scientific insight into the management of infrastructure asset. With the advent of the data age, the value of accessible data becomes increasingly evident. In SHM, many new data-centric methods are currently being developed at a high pace. A consequent application of research data management (RDM) concepts in SHM projects enables a systematic management of raw and processed data, and thus facilitates the development and application of artificial intelligence (AI) and machine learning (ML) methods to the SHM data. In this contribution, a case study based on an institutional RDM framework is presented. Data and metadata from monitoring the structural health of the Maintalbrücke Gemünden for a period of 16 months are managed with the RDM system BAM Data Store, which makes use of the openBIS data management software. An ML procedure is used to classify the data. Feature engineering, feature training and resulting data
are performed and modelled in the RDM system.
Applying data-driven AI systems makes it possible to extract patterns from given data, generate predictions and helps making decisions. Material research and testing holds a plethora of AI-based applications, for example, for the automatized search and synthesis of new materials, the detection of materials defects, or the prediction of process and materials parameters (inverse problems). However, AI algorithms can often only be as good as the training data from which the corresponding models are learned. Therefore, it is also indispensable to develop measures for the standardization and quality assurance of such data.
For this purpose, we develop and implement methods from transferring data from various sources into a homogeneous data repository with uniform data descriptions. Through the standardization and corresponding machine-readable interfaces, research data can be made usable and reusable for further data analyses. In addition to the technical implementation of integrative platforms, it is crucial that quality-assured research data management is recognized and implemented as an integral part of daily scientific work. Finally, we provide a vision of how the Federal Institute for Materials Research and Testing can benefit from data-driven AI systems. We discuss early applications and take a peek at future research.
The increasing adoption of Open Science principles has been a prevalent topic in the welding science community over the last years. Providing access to welding knowledge in the form of complex and complete datasets in addition to peer-reviewed publications can be identified as an important step to promote knowledge exchange and cooperation. There exist previous efforts on building data models specifically for fusion welding applications; however, a common agreed upon implementation that is used by the community is still lacking. One proven approach in other domains has been the use of an openly accessible and agreed upon file and data format used for archiving and sharing domain knowledge in the form of experimental data. Going into a similar direction, the welding community faces particular practical, technical, and also ideological challenges that are discussed in this paper. Collaboratively building upon previous work with modern tools and platforms, the authors motivate, propose, and outline the use of a common file format specifically tailored to the needs of the welding research community as a complement to other already established Open Science practices. Successfully establishing a culture of openly accessible research data has the potential to significantly stimulate progress in welding research.
Scientific welding data covers a wide range of physical domains and timescales and are measured using various different sensors. Complex and highly specialized experimental setups at different welding institutes complicate the exchange of welding research data further. The WelDX research project aims to foster the exchange of scientific data inside the welding community by developing and establishing a new open source file format suitable for the documentation of experimental welding data and upholding associated quality standards. In addition to fostering scientific collaboration inside the national and international welding community an associated advisory committee will be established to oversee the future development of the file format. The proposed file format will be developed with regard to current needs of the community regarding interoperability, data quality and performance and will be published under an appropriate open source license. By using the file format objectivity, comparability and reproducibility across different experimental setups can be improved.
The talk gives an introduction into gas metal arc welding and its relation to plasma science as well as current challenges in welding research concerning research data management and the application of the FAIR principles.
The WelDX project is introduced and the main goals are discussed and contrasted with the current features of the weldx API.
Different internal and public facing use cases focusing on research data management and their implementation using weldx are presented.
The interactive part of the presentation displays some advanced multi layer use cases and data analysis using the weldx API as well as the integration of materials properties into weldx.
The presentation shows the current development status of the Welding Data Exchange (weldx) format based on the Advanced Scientific Data Format (ASDF).
The use of a complete single pass arc welding experiment example stored in a single weldx file and validated against a predefined schema definition is presented.
The example includes generic experimental metadata, the workpiece geometry and materials definition following associated standards, the weld process spatial movement description, the welding process parameter descriptions and welding process measurements. The inclusion of 3D scan data of the workpiece description is also included.
The full code and data is available on GitHub: https://github.com/BAMWelDX/IIW2021_AA_CXII
The presentation shows the current development status of the Welding Data Exchange (weldx) format based on the Advanced Scientific Data Format (ASDF).
The use of a complete single pass arc welding experiment example stored in a single weldx file and validated against a predefined schema definition is presented.
The example includes generic experimental metadata, the workpiece geometry and materials definition following associated standards, the weld process spatial movement description, the welding process parameter descriptions and welding process measurements. The inclusion of 3D scan data of the workpiece description is also included.
The full code and data is available on GitHub: https://github.com/BAMWelDX/IIW2021_joint_intermediate_CXII
The talk motivates and introduces the WelDX project and the proposed solutions for current challenges in the field of research data management and Open Science practices in welding research.
Using an exemplary welding dataset based on the joint and welding process design of offshore structures, advanced data fusion and analysis capabilities are demonstrated.
The dataset shown consists of a complex welding sequence covering multiple weld layers with varying process parameters and adaptive weaving motions to cover manufacturing tolerances. In the presentation, an interactive exploration of the dataset contents in the spatial domain is presented. Furthermore, transformation between spatial and time domain of the data is demonstrated.
In addition to data gathered during the welding process, the integration of downstream testing data and results is also explained. For demonstration, integration of weld seam cross section images and Vickers hardness mapping test results into the dataset are explained an demonstrated. The testing data is set into context with the welding process information.
Finally, implications for advancements in research data management for WAAM and AI applications are discussed.
Scientific welding data covers a wide range of physical domains and timescales and are measured using various different sensors. Complex and highly specialized experimental setups at different welding institutes complicate the exchange of welding research data further. The project aims to simplify the exchange and quality assurance of scientific welding data by establishing a common open source file format.