The study explores vibrations, and thermal and tool wear (TW) mechanisms during dry end milling of AISI D2 steel employing two-, four-, and six- flute designs of TiAlN-PVD coated carbide end mill cutting tools. Because of multi-physics interactions in machining zone, supervised machine learning is used to understand TW, machining vibrations (MV), surface roughness (Ra) and the temperature of machining zone (TMZ), and to correlate the surface characteristics. A dedicated full factorial design of experiments campaign involving flutes (Fn), feed rate (Fr), and depth of cut (Doc) as variables is carried out to learn tool-workpiece interfacial characteristics. Results revealed that MV, Ra, TMZ and TW improved 57.82 %, 87.11 %, 79.22 % and 79.40 %, respectively at Fn = 6, Fr = 300 mm/min and Doc = 0.5 mm. Supervised machine learning (neural networks) and non-dominated sorting genetic algorithm (NSGA-II) based results showed reduction in vibrations by 89.84 %, Ra 97.39 %, tool-workpiece interaction temperature by 56.86 % and TW by 57.07 % by understanding mechanistic correlations of machining parameters with performance metrics.
Read full abstract