@inproceedings{LippelBihlmeierStuhlsatz2025, author = {Lippel, Jens and Bihlmeier, Richard and Stuhlsatz, Andr{\´e}}, title = {A Computer Vision Approach to Fertilizer Detection and Classification}, series = {Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications}, booktitle = {Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications}, publisher = {SCITEPRESS - Science and Technology Publications}, doi = {10.5220/0013189300003912}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-54735}, pages = {563 -- 569}, year = {2025}, subject = {Maschinelles Lernen}, language = {en} } @article{YuwonoHussainSchwungetal.2025, author = {Yuwono, Steve and Hussain, Ahmar Kamal and Schwung, Dorothea and Schwung, Andreas}, title = {Self-optimization in distributed manufacturing systems using Modular State-based Stackelberg games}, series = {Journal of Manufacturing Systems}, volume = {80}, journal = {Journal of Manufacturing Systems}, publisher = {Elsevier}, issn = {0278-6125}, doi = {10.1016/j.jmsy.2025.03.025}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-55471}, pages = {578 -- 594}, year = {2025}, abstract = {In this study, we introduce Modular State-based Stackelberg Games (Mod-SbSG), a novel game structure developed for distributed self-learning in modular manufacturing systems. Mod-SbSG enhances cooperative decision-making among self-learning agents within production systems by integrating State-based Potential Games (SbPG) with Stackelberg games. This hierarchical structure assigns more important modules of the manufacturing system a first-mover advantage, while less important modules respond optimally to the leaders' decisions. This decision-making process differs from typical multi-agent learning algorithms in manufacturing systems, where decisions are made simultaneously. We provide convergence guarantees for the novel game structure and design learning algorithms to account for the hierarchical game structure. We further analyse the effects of single-leader/multiple-follower and multiple-leader/multiple-follower scenarios within a Mod-SbSG. To assess its effectiveness, we implement and test Mod-SbSG in an industrial control setting using two laboratory-scale testbeds featuring sequential and serial parallel processes. The proposed approach delivers promising results compared to the vanilla SbPG, which reduces overflow by 97.1\%, and in some cases, prevents overflow entirely. Additionally, it decreases power consumption by 5\%-13\% while satisfying the production demand, which significantly improves potential (global objective) values.}, subject = {Intelligente Fertigung}, language = {en} } @inproceedings{KuehnelEissmannWiedeetal.2024, author = {K{\"u}hnel, Johannes and Eißmann, Timo and Wiede, Christian and Schwung, Dorothea and Grabmaier, Anton}, title = {Semi-Supervised Anomaly Detection in the TinyML Domain Through Multi-Target Few-Shot Domain Adaptation}, series = {2024 IEEE 29th International Conference on Emerging Technologies and Factory Automation (ETFA)}, volume = {133}, booktitle = {2024 IEEE 29th International Conference on Emerging Technologies and Factory Automation (ETFA)}, publisher = {IEEE}, doi = {10.1109/ETFA61755.2024.10710811}, pages = {8}, year = {2024}, subject = {Maschinelles Lernen}, language = {en} } @unpublished{YuwonoSchwungSchwung2024, author = {Yuwono, Steve and Schwung, Dorothea and Schwung, Andreas}, title = {Transfer learning of state-based potential games for process optimization in decentralized manufacturing systems}, series = {arXiv}, journal = {arXiv}, publisher = {arXiv}, doi = {10.48550/arXiv.2408.05992}, pages = {48}, year = {2024}, abstract = {This paper presents a novel transfer learning approach in state-based potential games (TL-SbPGs) for enhancing distributed self-optimization in manufacturing systems. The approach focuses on the practical relevant industrial setting where sharing and transferring gained knowledge among similar-behaved players improves the self-learning mechanism in large-scale systems. With TL-SbPGs, the gained knowledge can be reused by other players to optimize their policies, thereby improving the learning outcomes of the players and accelerating the learning process. To accomplish this goal, we develop transfer learning concepts and similarity criteria for players, which offer two distinct settings: (a) predefined similarities between players and (b) dynamically inferred similarities between players during training. We formally prove the applicability of the SbPG framework in transfer learning. Additionally, we introduce an efficient method to determine the optimal timing and weighting of the transfer learning procedure during the training phase. Through experiments on a laboratory-scale testbed, we demonstrate that TL-SbPGs significantly boost production efficiency while reducing power consumption of the production schedules while also outperforming native SbPGs.}, subject = {Maschinelles Lernen}, language = {en} } @unpublished{YuwonoLoeppenbergSchwungetal.2024, author = {Yuwono, Steve and L{\"o}ppenberg, Marlon and Schwung, Dorothea and Schwung, Andreas}, title = {Gradient-based Learning in State-based Potential Games for Self-Learning Production Systems}, series = {arxiv}, journal = {arxiv}, publisher = {arxiv}, doi = {10.48550/arXiv.2406.10015}, year = {2024}, abstract = {In this paper, we introduce novel gradient-based optimization methods for state-based potential games (SbPGs) within self-learning distributed production systems. SbPGs are recognised for their efficacy in enabling self-optimizing distributed multi-agent systems and offer a proven convergence guarantee, which facilitates collaborative player efforts towards global objectives. Our study strives to replace conventional ad-hoc random exploration-based learning in SbPGs with contemporary gradient-based approaches, which aim for faster convergence and smoother exploration dynamics, thereby shortening training duration while upholding the efficacy of SbPGs. Moreover, we propose three distinct variants for estimating the objective function of gradient-based learning, each developed to suit the unique characteristics of the systems under consideration. To validate our methodology, we apply it to a laboratory testbed, namely Bulk Good Laboratory Plant, which represents a smart and flexible distributed multi-agent production system. The incorporation of gradient-based learning in SbPGs reduces training times and achieves more optimal policies than its baseline.}, subject = {Maschinelles Lernen}, language = {en} } @article{BerkantOezelDurmazNurOezetal.2025, author = {Berkant {\"O}zel, Mehmet and Durmaz, Ufuk and Nur {\"O}z, Muhammed Ali and {\"U}mit Tepe, Ahmet and {\"O}z, Cemil and Uysal, {\"U}nal and Yal{\c{c}}inkaya, Orhan and Cemal Beni̇m, Ali and Alomayrah, Norah and Al-Buriahi, M.S.}, title = {Enhanced prediction of heat transfer in jet impingement cooling using an artificial intelligence: A case study}, series = {Case Studies in Thermal Engineering}, volume = {73}, journal = {Case Studies in Thermal Engineering}, publisher = {Elsevier}, issn = {2214-157X}, doi = {10.1016/j.csite.2025.106605}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-55017}, pages = {11}, year = {2025}, abstract = {This paper is about the gas turbine cooling model using the extended jet array. The aim of the present study is to explain the effect of the impinging jet configuration of elongated jet holes on the heat transfer distribution with a deep Convolutional Neural Network (CNN). CNNs preserve spatial relationships through convolutional operations—unlike traditional Artificial Neural Networks (ANNs), which require flattening of input data—allowing for higher accuracy with smaller datasets and faster training due to fewer trainable parameters. The goal is to understand how accurate and fast deep learning models with limited data can deliver predictions for complex systems such as jet impingement cooling. The proposed CNN model was compared with the numerical and experimental data. Jet impingement cooling was examined with four different Reynolds numbers (16250, 21700, 27100, 36250) and six dimensionless gaps between the jet and the target surface (G/D = 1, 2, 3, 4, 5, and 6). Mean and local Nusselt number (Nu) distributions were investigated in particular. G/D = 2 was selected as the test data for the model among the configurations. While the CFD model failed to identify this peak performance point, the CNN model accurately captured this maximum behavior despite being trained only on other configurations. Moreover, the CNN model demonstrated more suitable prediction accuracy in the first two jet impingement regions than the CFD data, especially in capturing peak Nusselt numbers and localized hot spots. It has been determined that the CNN model provided more accurate predictions of the local Nusselt numbers within the measurement region compared to the Computational Fluid Dynamics (CFD) study for the average Nu, the proposed CNN model showed a closer agreement to experimental data than numerical simulation by 4.15 \% at the Re = 32500 and G/D = 2.}, subject = {Maschinelles Lernen}, language = {en} } @unpublished{YuwonoSchwungSchwung2024, author = {Yuwono, Steve and Schwung, Dorothea and Schwung, Andreas}, title = {Distributed Stackelberg Strategies in State-based Potential Games for Autonomous Decentralized Learning Manufacturing Systems}, publisher = {arXiv}, doi = {10.48550/arXiv.2408.06397}, pages = {14}, year = {2024}, abstract = {This article describes a novel game structure for autonomously optimizing decentralized manufacturing systems with multi-objective optimization challenges, namely Distributed Stackelberg Strategies in State-Based Potential Games (DS2-SbPG). DS2-SbPG integrates potential games and Stackelberg games, which improves the cooperative trade-off capabilities of potential games and the multi-objective optimization handling by Stackelberg games. Notably, all training procedures remain conducted in a fully distributed manner. DS2-SbPG offers a promising solution to finding optimal trade-offs between objectives by eliminating the complexities of setting up combined objective optimization functions for individual players in self- learning domains, particularly in real-world industrial settings with diverse and numerous objectives between the sub-systems. We further prove that DS2-SbPG constitutes a dynamic potential game that results in corresponding converge guarantees. Experimental validation conducted on a laboratory-scale testbed highlights the efficacy of DS2-SbPG and its two variants, such as DS2-SbPG for single-leader-follower and Stack DS2-SbPG for multi-leader-follower. The results show significant reductions in power consumption and improvements in overall performance, which signals the potential of DS2-SbPG in real-world applications.}, subject = {Maschinelles Lernen}, language = {en} } @article{MandalGuptaBiswasetal.2025, author = {Mandal, Dipak Kumar and Gupta, Kritesh Kumar and Biswas, Nirmalendu and Manna, Nirmal K. and Santra, Somnath and Benim, Ali Cemal}, title = {Optimization of hybrid solar chimney power plants (HSCPPs): A review of multi-objective approaches}, series = {Applied Energy}, volume = {396}, journal = {Applied Energy}, publisher = {Elsevier}, issn = {0306-2619}, doi = {10.1016/j.apenergy.2025.126214}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-54199}, pages = {25}, year = {2025}, subject = {Erneuerbare Energien}, language = {en} } @inproceedings{KerstenHaaseRiess2025, author = {Kersten, Thierry and Haase, Thomas and Riess, Bernhard M.}, title = {Evaluation of a new AI driven SoC Design Flow Approach}, series = {MPC-Workshop Januar 2024 / Januar 2025}, volume = {67/68}, booktitle = {MPC-Workshop Januar 2024 / Januar 2025}, editor = {Schmidt, Lothar}, publisher = {Technische Hochschule Ulm}, address = {Ulm}, organization = {Multi Projekt Chip Gruppe}, issn = {1868-9221}, pages = {47 -- 57}, year = {2025}, subject = {K{\"u}nstliche Intelligenz}, language = {en} } @techreport{PolklasWillekeLehmbrincketal.2025, author = {Polklas, Thomas and Willeke, Tobias and Lehmbrinck, Marc and Fomina, Anastasia and Maqueo Martinez, Manuel Ernesto and Oliveira Brito, Marcel}, title = {RoboFlexRobuste Turbomaschinen f{\"u}r den flexiblen Einsatz: Numerische und experimentelle Untersuchungen zu einem axial-radial Turbinendiffusor (AP 1.5); KI-basierte, modellpr{\"a}diktive Turbostrang-Regelung f{\"u}r einen energieoptimalen Betrieb (AP 4.2) (Abschlussbericht Verbundprojekt AG Turbo Teilvorhaben-Nr. 1.5 und 4.2)}, publisher = {Repositorium f{\"u}r Naturwissenschaften und Technik}, address = {Hannover}, doi = {10.34657/18643}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-57122}, pages = {137}, year = {2025}, abstract = {In der Zusammenarbeit zwischen dem ITSM der Universit{\"a}t Stuttgart und Everllence wurde im Rahmen dieses Vorhabens (AP 1.5) eine Versuchsanlage zur Erzeugung repr{\"a}sentativer Diffusorstr{\"o}mungen entwickelt und f{\"u}r experimentelle Untersuchungen eingesetzt. Begleitend zu den experimentellen Untersuchungen wurden verschiedene numerische Modelle entwickelt und anhand der Messergebnisse validiert. Aus diesem Vergleich wurden Empfehlungen zur numerischen Str{\"o}mungssimulation von Axial-Radial-Diffusoren abgeleitet. Insgesamt zeigte die numerische Simulation unter den gew{\"a}hlten Bedingungen eine hohe Sensitivit{\"a}t gegen{\"u}ber den Modellparametern mit entsprechenden Auswirkungen auf die beabsichtigte Geometrieoptimierung. In Arbeitspaket 4.2 des AG Turbo Verbundprojekts RoboFlex ist es den Projektpartnern Hochschule D{\"u}sseldorf und Everllence gelungen, eine modellpr{\"a}diktive Mehrgr{\"o}ßen-Verbundregelung eines exemplarischen Turbomaschinen-Strangs zu entwickeln. Dies erlaubt der Regelungseinheit, alle verf{\"u}gbaren Stellgr{\"o}ßen simultan zu benutzen, um ein vorgegebenes Regelungsziel zu erreichen. Zudem kann das Regelungsziel flexibel um weitere Vorgaben und Beschr{\"a}nkungen erweitert werden, indem diese in die Kostenfunktion implementiert werden. Ein zentraler Forschungsschwerpunkt war hierbei die Einbindung von Methoden der k{\"u}nstlichen Intelligenz in die Modellierung und im gleichen Zuge deren {\"U}berf{\"u}hrung und Implementierung in die linearisierte Modellform, die die modellpr{\"a}diktive Regelung verlangt. Ein besonderes Augenmerk wurde auf die fortlaufende Nachf{\"u}hrung des KI-Modells an den zuletzt durch Messungen bekannten Anlagenzustand gelegt.}, subject = {Str{\"o}mungsmaschine}, language = {de} }