Industrial Deep Learning
In the field of "Industrial Deep Learning", we research deep learning technologies for industrial applications in order to realise innovative solutions in production, logistics and the environment. We combine basic AI research with industrial practice and focus on three main areas
Image-based methods for the automation of quality controls and the precise localisation of anomalies and damage.
Sensor-based situation and condition assessment
Processing and utilisation of transient and seasonal sensor data for condition monitoring, anomaly detection and forecasting.
Intelligent planning and process design
Learning methods for solving complex planning and optimisation problems and for evaluating and parameterising processes.
Our research addresses a broad spectrum of deep learning technologies, including different learning paradigms such as supervised and reinforcement learning, learning scenarios such as transfer learning, representation learning and explainable AI, as well as model architectures such as transformer networks, autoencoders and generative adversarial networks.
We work closely with industry partners, whether as part of publicly funded projects or direct R&D contracts. In doing so, we deal intensively with real challenges and always take the needs of end users and technical experts into account. This practical orientation ensures that our research results are not only theoretically sound, but also directly applicable in industrial practice and improve value creation.
Selected publications
- 2023
- Bulow, F., & Meisen, T. (2023). "A review on methods for state of health forecasting of lithium-ion batteries applicable in real-world operational conditions" , Journal of Energy Storage , 57 , 105978.
- Waubert-de-Puiseau, C., Tercan, H., & Meisen, T. (2023). "Curriculum Learning in Job Shop Scheduling using Reinforcement Learning" .
- Samsonov, V., Chrismarie, E., Köpken, H., Bär, S., Lütticke, D., & Meisen, T. (2023). "Deep representation learning and reinforcement learning for workpiece setup optimization in CNC milling" , Production Engineering .
- Puiseau, C., Zey, L., Demir, M., Tercan, H., & Meisen, T. (2023). "On The Effectiveness Of Bottleneck Information For Solving Job Shop Scheduling Problems Using Deep Reinforcement Learning" , Proceedings of the Conference on Production Systems and Logistics: CPSL 2023-2 .
- Waubert-de-Puiseau, C., Peters, J., Dörpelkus, C., Tercan, H., & Meisen, T. (2023). "schlably: A Python Framework for Deep Reinforcement Learning Based Scheduling Experiments" , arXiv arXiv:2301.04182 .
