The concept of Self Organizing Maps for anomilies detection can be used to map appearing loads in the simulation on experimental loads indicating if it is sufficiently covered by experiment. We often use these methods to check the validity of machine learning models. In addition to the model quality, it is also particularly important to know the model validity limits.

• Check if the calibrated models are valid for the loadcases which occur in the simulation.
• Capable to find new load patterns

see also

• A. Barolome, T. Palau, A. Kuhn, A. Rauh, H.U. Mader: A Machine Learning based Approach to large scale Material Model Evaluation and iterative experimental design, European Congress on Computational Methods in Applied Sciences and Engineering (ECCOMAS 2012), Vienna, Austria, September 10-14, 2012