In current research on simulation-driven product development, machine learning surrogate models (MLS) have been introduced to improve or partially replace traditional simulations for product analysis and validation. MLS techniques significantly reduce the computational effort compared to traditional simulations. In this way, a large solution space can be examined with little effort.

At the same time, the need for automation in product manufacturing is growing, not least due to the demographically-related shortage of skilled workers. In this context, Design for Automatic Assembly methods have become established.

The EasyMLS research project combines the use of MLS and the need for hybrid decision-making. The desired results form the methodological basis for the implementation of corresponding assistance functions in CAD and other digital engineering tools to increase efficiency in product development.

Aims of the project

• Apply MLS approaches to late-stage product development analysis in early design stages
• Enable the proactive reduction of effort in later phases of product development
• Significantly reduce computational costs through the use of surrogate models that enable direct feedback in engineering tools
• An early prediction and adjustment of the product with regard to its suitability for automated assembly

Focus of the project

• Optimization of the spatial arrangement of product components
• Enabling automated assembly approaches by adjusting the assembly sequence
• Combining both approaches into a single, lean process
• Integration of the researched process into engineering software
• Validation of the approach using an electrolyzer