Antonia Wüst
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Antonia Wüst has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Jonas Moske
Institute for Production Engineering and Forming Machines (PtU), Technical University of Darmstadt, Darmstadt, Germany
7 shared publications - Markus Schumann
Institute for Production Engineering and Forming Machines (PtU), Technical University of Darmstadt, Darmstadt, Germany
7 shared publications - Kristian Kersting
Technical University of Darmstadt
6 shared publications - Felix Divo
AIML Lab, Technical University Darmstadt, Darmstadt, Germany
4 shared publications - Peter Groche
Institute for Production Engineering and Forming Machines, Technical University of Darmstadt, Otto-Berndt-Straße 2, 64287 Darmstadt, Germany
4 shared publications - Peter Groche
TU Darmstadt
3 shared publications
Latest Publications
- Open Access
- 28 May 2026
- Published by Springer Nature in Transactions of the Indian Institute of Metals
- Vol. 79 (6) , 140
Information Content Analysis of Direct and Indirect Force Measurements for Machine Learning-Based Process State Classification in Multi-stage Sheet Metal Forming
The predictive power of machine learning models for process monitoring in sheet metal forming depends strongly on the information content of the sensor signals. This study investigates how force signal characteristics represent process conditions in a multi-stage forming process consisting of deep drawing and ironing, in which surface roughness evolves with a downstream tendency. Indirect and direct force measurement concepts are compared: While indirect sensors are prone to noise, direct sensors often show more clarity. Neural networks are trained on datasets covering multiple roughness and process configurations. Model performance is analysed using classification metrics and explainable AI methods. The results reveal a counter-intuitive finding: Visually smooth force signals with high signal-to-noise ratio can provide limited or misleading information for convolutional neural networks due to temporal misalignment, whereas noisier signals with distributed dynamics show more robust predictions. The study shows the influence of signal clarity for data-driven process monitoring in Industry 4.0-enabled forming.
- 28 May 2026
- Published by Springer Nature in Transactions of the Indian Institute of Metals
- Vol. 79 (6) , 140
Open AccessInformation Content Analysis of Direct and Indirect Force Measurements for Machine Learning-Based Process State Classification in Multi-stage Sheet Metal Forming
The predictive power of machine learning models for process monitoring in sheet metal forming depends strongly on the information content of the sensor signals. This study investigates how force signal characteristics represent process conditions in a multi-stage forming process consisting of deep drawing and ironing, in which surface roughness evolves with a downstream tendency. Indirect and direct force measurement concepts are compared: While indirect sensors are prone to noise, direct sensors often show more clarity. Neural networks are trained on datasets covering multiple roughness and process configurations. Model performance is analysed using classification metrics and explainable AI methods. The results reveal a counter-intuitive finding: Visually smooth force signals with high signal-to-noise ratio can provide limited or misleading information for convolutional neural networks due to temporal misalignment, whereas noisier signals with distributed dynamics show more robust predictions. The study shows the influence of signal clarity for data-driven process monitoring in Industry 4.0-enabled forming.
- Open Access
- 7 May 2026
- Published by Springer Nature in Production Engineering
- Vol. 20 (2) , 66
Structured representation of simulation and annotation data for machine learning in forming technologies
The use of machine learning (ML) in manufacturing requires structured, especially standardized, access to both simulation data and domain knowledge. This paper introduces a JSON-based data format for representing synthetic force-time series alongside expert annotations. The schema captures simulation metadata, tool and material parameters, and allows explicit expert knowledge, such as failure indicators, to be linked to signal segments. The proposed structure enables process-aware ML methods that leverage both domain knowledge and raw data for improved learning and generalization. A deep drawing use case illustrates how the format facilitates knowledge-guided learning. The approach aims to bridge the gap between real and simulated production data, supporting scalable integration in modern manufacturing systems.
- 7 May 2026
- Published by Springer Nature in Production Engineering
- Vol. 20 (2) , 66
Open AccessStructured representation of simulation and annotation data for machine learning in forming technologies
The use of machine learning (ML) in manufacturing requires structured, especially standardized, access to both simulation data and domain knowledge. This paper introduces a JSON-based data format for representing synthetic force-time series alongside expert annotations. The schema captures simulation metadata, tool and material parameters, and allows explicit expert knowledge, such as failure indicators, to be linked to signal segments. The proposed structure enables process-aware ML methods that leverage both domain knowledge and raw data for improved learning and generalization. A deep drawing use case illustrates how the format facilitates knowledge-guided learning. The approach aims to bridge the gap between real and simulated production data, supporting scalable integration in modern manufacturing systems.
- Open Access
- 30 March 2026
- Published by Springer Nature in Production Engineering
- Vol. 20 (2) , 61
Camera-based feature extraction and uncertainty analysis in deep drawing in progressive dies
In progressive die deep drawing, small deviations in strip positioning and transport can accumulate across stages, leading to significant geometric variations and quality issues. Reliable datasets are therefore essential for data-driven modeling, yet process labels such as the nominal feed length are often affected by machine tolerances and thus introduce uncertainty. This work presents a camera-based methodology for automated feature extraction at the end of the deep drawing process. Using an inline profile projector, relevant geometric features such as circle distances, centroid positions, and local material widths are extracted from high-speed image data. The implemented algorithm combines edge detection, circle fitting, centroid analysis, and directional width measurements to generate consistent feature sets under industrial conditions. Challenges such as material reflections, asymmetric deformation, and non-ideal boundary conditions are addressed, and geometry-specific obstacles for feature detection are discussed. The extracted features are further analyzed with respect to their statistical variability across systematically varied feed length values. Results demonstrate that the variance of image-based features increases with higher feed length, indicating that nominal process labels do not always reflect the actual material position. The study highlights the importance of uncertainty quantification in dataset generation for machine learning in forming technology. By linking process variation to directly observable geometry features, the proposed approach provides both methodological guidance for feature extraction and conceptual insights into label quality in industrial datasets.
- 30 March 2026
- Published by Springer Nature in Production Engineering
- Vol. 20 (2) , 61
Open AccessCamera-based feature extraction and uncertainty analysis in deep drawing in progressive dies
In progressive die deep drawing, small deviations in strip positioning and transport can accumulate across stages, leading to significant geometric variations and quality issues. Reliable datasets are therefore essential for data-driven modeling, yet process labels such as the nominal feed length are often affected by machine tolerances and thus introduce uncertainty. This work presents a camera-based methodology for automated feature extraction at the end of the deep drawing process. Using an inline profile projector, relevant geometric features such as circle distances, centroid positions, and local material widths are extracted from high-speed image data. The implemented algorithm combines edge detection, circle fitting, centroid analysis, and directional width measurements to generate consistent feature sets under industrial conditions. Challenges such as material reflections, asymmetric deformation, and non-ideal boundary conditions are addressed, and geometry-specific obstacles for feature detection are discussed. The extracted features are further analyzed with respect to their statistical variability across systematically varied feed length values. Results demonstrate that the variance of image-based features increases with higher feed length, indicating that nominal process labels do not always reflect the actual material position. The study highlights the importance of uncertainty quantification in dataset generation for machine learning in forming technology. By linking process variation to directly observable geometry features, the proposed approach provides both methodological guidance for feature extraction and conceptual insights into label quality in industrial datasets.
- Open Access
- 28 March 2026
- Published by Springer Nature in Research Square
Information Content Analysis of Direct and Indirect Force Measurements for Machine Learning-Based Process State Classification in Multi-Stage Sheet Metal Forming
The predictive power of machine learning models for process monitoring in sheet metal forming depends strongly on the information content of the sensor signals. This study investigates how force signal characteristics represent process conditions in a multi-stage forming process consisting of deep drawing and ironing, in which surface roughness evolves with a downstream tendency. Indirect and direct force measurement concepts are compared: While indirect sensors are prone to noise, direct sensors often show more clarity. Neural networks are trained on datasets covering multiple roughness and process configurations. Model performance is analysed using classification metrics and explainable AI methods. The results reveal a counterintuitive finding: Visually smooth force signals with high signal-to-noise ratio can provide limited or misleading information for convolutional neural networks due to temporal misalignment, whereas noisier signals with distributed dynamics show more robust predictions. The study shows the influence of signal clarity for data-driven process monitoring in Industry 4.0-enabled forming.
- 28 March 2026
- Published by Springer Nature in Research Square
Open AccessInformation Content Analysis of Direct and Indirect Force Measurements for Machine Learning-Based Process State Classification in Multi-Stage Sheet Metal Forming
The predictive power of machine learning models for process monitoring in sheet metal forming depends strongly on the information content of the sensor signals. This study investigates how force signal characteristics represent process conditions in a multi-stage forming process consisting of deep drawing and ironing, in which surface roughness evolves with a downstream tendency. Indirect and direct force measurement concepts are compared: While indirect sensors are prone to noise, direct sensors often show more clarity. Neural networks are trained on datasets covering multiple roughness and process configurations. Model performance is analysed using classification metrics and explainable AI methods. The results reveal a counterintuitive finding: Visually smooth force signals with high signal-to-noise ratio can provide limited or misleading information for convolutional neural networks due to temporal misalignment, whereas noisier signals with distributed dynamics show more robust predictions. The study shows the influence of signal clarity for data-driven process monitoring in Industry 4.0-enabled forming.
- Open Access
- 13 January 2026
- Published by Springer Nature in Research Square
Structured Representation of Simulation and Annotation Data for Machine Learning in Forming Technologies
The use of machine learning (ML) in manufacturing requires structured, especially standardized, access to both simulation data and domain knowledge. This paper introduces a JSON-based data format for representing synthetic force-time series alongside expert annotations. The schema captures simulation metadata, tool and material parameters, and allows explicit expert knowledge, such as failure indicators, to be linked to signal segments.The proposed structure enables process-aware ML methods that leverage both domain knowledge and raw data for improved learning and generalization. A deep drawing use case illustrates how the format facilitates knowledge-guided learning. The approach aims to bridge the gap between real and simulated production data, supporting scalable integration in modern manufacturing systems.
- 13 January 2026
- Published by Springer Nature in Research Square
Open AccessStructured Representation of Simulation and Annotation Data for Machine Learning in Forming Technologies
The use of machine learning (ML) in manufacturing requires structured, especially standardized, access to both simulation data and domain knowledge. This paper introduces a JSON-based data format for representing synthetic force-time series alongside expert annotations. The schema captures simulation metadata, tool and material parameters, and allows explicit expert knowledge, such as failure indicators, to be linked to signal segments.The proposed structure enables process-aware ML methods that leverage both domain knowledge and raw data for improved learning and generalization. A deep drawing use case illustrates how the format facilitates knowledge-guided learning. The approach aims to bridge the gap between real and simulated production data, supporting scalable integration in modern manufacturing systems.
- Open Access
- 21 November 2025
- Published by Springer Nature in Research Square
Requirements for numeric models as sources of synthetic data for predicting real-world data sets
In the field of forming technology, synthetic data generated by finite element (FE) simulations is increasingly being used to train machine learning (ML) models for quality prediction. However, the predictive accuracy on real-world process data is often limited by the so-called “reality gap” between simulated and measured signals. This study investigates how simulation model complexity influences the suitability of synthetic data for training ML models that generalize to real progressive deep drawing processes. Three representative simulation configurations of increasing complexity (L1–L3) are implemented and evaluated against experimental data collected under production conditions using sensorintegrated tools. The analysis covers multiple ML tasks, including classification of pre-process connector cut geometries, detection of process disturbances, separation of subtle geometry variants, assessment of feature transfer robustness, and saliency-based interpretation of signal regions. The results show that simple models (L1) enable robust classification of failures such as material damage within synthetic domains though their transferability to real data is restricted. Although the are computationally more expensive, higher complexity level (L2 and L3) better capture the effects of pre-processing and deformation-history, improving domain alignment and supporting physically meaningful model interpretation. Saliency map analysis reveals that models trained on synthetic data emphasize different signal regions than models trained on real data. This underscores the importance of task-relevant signal fidelity. The findings provide quantitative guidance for selecting an adequate level of simulation complexity and demonstrate that the underlying principles apply across a broader spectrum of modelling configurations.
- 21 November 2025
- Published by Springer Nature in Research Square
Open AccessRequirements for numeric models as sources of synthetic data for predicting real-world data sets
In the field of forming technology, synthetic data generated by finite element (FE) simulations is increasingly being used to train machine learning (ML) models for quality prediction. However, the predictive accuracy on real-world process data is often limited by the so-called “reality gap” between simulated and measured signals. This study investigates how simulation model complexity influences the suitability of synthetic data for training ML models that generalize to real progressive deep drawing processes. Three representative simulation configurations of increasing complexity (L1–L3) are implemented and evaluated against experimental data collected under production conditions using sensorintegrated tools. The analysis covers multiple ML tasks, including classification of pre-process connector cut geometries, detection of process disturbances, separation of subtle geometry variants, assessment of feature transfer robustness, and saliency-based interpretation of signal regions. The results show that simple models (L1) enable robust classification of failures such as material damage within synthetic domains though their transferability to real data is restricted. Although the are computationally more expensive, higher complexity level (L2 and L3) better capture the effects of pre-processing and deformation-history, improving domain alignment and supporting physically meaningful model interpretation. Saliency map analysis reveals that models trained on synthetic data emphasize different signal regions than models trained on real data. This underscores the importance of task-relevant signal fidelity. The findings provide quantitative guidance for selecting an adequate level of simulation complexity and demonstrate that the underlying principles apply across a broader spectrum of modelling configurations.
- Open Access
- 30 October 2025
- Published by Springer Nature in Research Square
Camera-based feature extraction and uncertainty analysis in deep drawing in progressive dies
In progressive die deep drawing, small deviations in strip positioning and transport can accumulate across stages, leading to significant geometric variations and quality issues. Reliable datasets are therefore essential for data-driven modeling, yet process labels such as the nominal infeed are often affected by machine tolerances and thus introduce uncertainty. This work presents a camera-based methodology for automated feature extraction at the end of the deep drawing process. Using an inline profile projector, relevant geometric features such as circle distances, centroid positions, and local material widths are extracted from high-speed image data. The implemented algorithm combines edge detection, circle fitting, centroid analysis, and directional width measurements to generate consistent feature sets under industrial conditions. Challenges such as material reflections, asymmetric deformation, and non-ideal boundary conditions are addressed, and geometry-specific obstacles for feature detection are discussed. The extracted features are further analyzed with respect to their statistical variability across systematically varied infeed values. Results demonstrate that the variance of image-based features increases with higher infeed, indicating that nominal process labels do not always reflect the actual material position. The study highlights the importance of uncertainty quantification in dataset generation for machine learning in forming technology. By linking process variation to directly observable geometry features, the proposed approach provides both methodological guidance for feature extraction and conceptual insights into label quality in industrial datasets.
- 30 October 2025
- Published by Springer Nature in Research Square
Open AccessCamera-based feature extraction and uncertainty analysis in deep drawing in progressive dies
In progressive die deep drawing, small deviations in strip positioning and transport can accumulate across stages, leading to significant geometric variations and quality issues. Reliable datasets are therefore essential for data-driven modeling, yet process labels such as the nominal infeed are often affected by machine tolerances and thus introduce uncertainty. This work presents a camera-based methodology for automated feature extraction at the end of the deep drawing process. Using an inline profile projector, relevant geometric features such as circle distances, centroid positions, and local material widths are extracted from high-speed image data. The implemented algorithm combines edge detection, circle fitting, centroid analysis, and directional width measurements to generate consistent feature sets under industrial conditions. Challenges such as material reflections, asymmetric deformation, and non-ideal boundary conditions are addressed, and geometry-specific obstacles for feature detection are discussed. The extracted features are further analyzed with respect to their statistical variability across systematically varied infeed values. Results demonstrate that the variance of image-based features increases with higher infeed, indicating that nominal process labels do not always reflect the actual material position. The study highlights the importance of uncertainty quantification in dataset generation for machine learning in forming technology. By linking process variation to directly observable geometry features, the proposed approach provides both methodological guidance for feature extraction and conceptual insights into label quality in industrial datasets.