Case Study: The University of Sheffield achieves 80% fewer composite drilling experiments with Intellegens Alchemite

Tooling optimisation for composite drilling using deep learning

The University of Sheffield Advanced Manufacturing Research Centre (AMRC) needed a better way to identify optimal cutting parameters for drilling fibre-reinforced polymer composites, where surface delamination causes major part rejection and traditional testing requires extensive experimentation. Intellegens’ Alchemite™ deep learning software was used to address this challenge, especially given the sparse dataset with 82% of target data missing.

Intellegens trained an Alchemite™ model on tooling time series data from 55 drill/composite pairs and 23 machining responses, using preprocessing and aggregation to improve performance. The model achieved an R-squared of 0.73 and showed that just 200 tests could provide strong predictive insight, enabling up to an 80% reduction in direct testing costs and helping identify irrelevant variables for further experimental streamlining.

The University of Sheffield