Quantification and minimization of uncertainty for guided wave-based structural health monitoring with artificial neural network approaches

• Applicants: Professor Dr.-Ing. Steffen Freitag; Dr.-Ing. Jochen Moll
• Subject Area: Applied Mechanics, Statics and Dynamics
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 566880366

The focus of the research project is to develop guided wave-based structural health monitoring (SHM) approaches with artificial neural networks for damage detection, damage localization as well as the prediction of damage type and size. Feedforward and convolutional neural network architectures will be investigated within classification and regression tasks. For the training and validation of these machine learning approaches, experimental data as well as numerical simulation data of carbon-fiber composite plate structure stiffened by an omega stringer will be used. The objective of the research project is to quantify and to minimize the uncertainty within guided wave-based SHM in combination with artificial neural networks in order to get reliable and robust predictions, which is important to set up AI-based SHM systems. The following research questions will be answered within the project: How can physical knowledge of guided wave propagation be considered within artificial neural networks for damage detection, damage localization, damage type and size prediction? What are the key parameters to minimize the uncertainty in AI-based guided wave SHM systems? How can measured data and data from numerical simulations be fused to improve the reliability of AI-based guided wave SHM systems?

• DFG Programme: Research Grants
• International Connection: Italy
• Cooperation Partner: Dr.-Ing. Vittorio Memmolo