The research project "Analysis of the trustworthiness of models on the example of machining for qualification for the higher-level planning of manufacturing processes" is a cooperation project between the Chair of Data Science and Data Engineering, the Software Engineering by Algorithms and Logic working group and the Virtual Machining working group within the Industry and Production research area of the Lamar Institute for Machine Learning and Artificial Intelligence. The aim of the project is to establish a scientific foundation in order to improve the accuracy and reliability of data-based models for predicting process characteristics in machining production.

A major focus is the research of hybrid learning methods that combine simulations and real measurements. These approaches contribute to minimising the number of necessary real experiments by generating additional data through simulations. In addition, the extension of the feature space through the simulation of non-measurable process quantities is being investigated in order to further improve the accuracy of predictions. Furthermore, generative modelling approaches will be used to augment the database and thus increase resource efficiency.

To increase the interpretability and acceptance of data-based models in industry, the integration of explainability methods from the field of trustworthy AI will also be explored. This is particularly important in order to enable the application of the models in safety-critical applications.