Machine learning based track height prediction for complex tool paths in direct metal deposition
Daniel Knüttel, Stefano Baraldo, Anna Valente, Friedrich Bleicher, Konrad Wegener, Emanuele Carpanzano
Abstract
The tooling industry persistently demands for advanced techniques to boost the tools performances over their lifecycle. Direct Metal Deposition (DMD) presents key opportunities in the tool refurbishment. However, the typical tool paths via DMD consist of alternated smooth segments and sharp corners. Here, the fluctuation of energy density and powder quantities often cause critical geometrical deviations to the tool restored sections. This work presents a novel machine learning based prediction approach that characterizes paths using features associated to process parameters and performed geometry. The benefits of the approach have been validated on toolpaths, which typically characterize a tool refurbishment process.