Multi-response optimization and machine learning-based prediction of straight-groove warm incremental sheet forming...

This study investigates the warm straight-groove incremental sheet forming (ISF) behavior of AZ31 magnesium alloy using an integrated experimental, statistical, and machine learning approach. To test the effect of forming temperature, step-down, spindle speed and feed rate, a Taguchi L27 design was used to study the effect of above variables on forming time and forming force. TOPSIS multi-response optimization was used to find the most balanced parameter combination to result in low force and high process efficiency. The statistical result showed that temperature and step-down were the most prevailing factors that controlled the deformation behaviour at warm forming conditions. A Random Forest regression model was constructed in order to increase the predictive ability, and it was able to successfully recreate the trends in the forming time, forming force, and performance index. The fractographic analysis of the fractured wall of the groove proved the presence of a ductile failure mechanism in which voids and localisation of shear dominate. The combined DOE-TOPSIS-ML-SEM analysis offers a very powerful procedure of comprehending and optimizing the warm incremental sheet forming of lightweight AZ31 magnesium alloy.