Machine Learning and NSGA-II Integration for Optimization in AISI 52100 Steel Turning

This work proposes a methodology for multiobjective optimization of the cutting parameters in the turning of hardened AISI 52100 steel, aiming to balance the reduction of machining time and the increase of tool life. Using vibration signals collected during the tests, relevant characteristics were extracted and several Machine Learning (ML) algorithms were employed, combined in an ensemble stacking model to predict tool wear (VBc). The predictive model, validated with high accuracy (R² = 0.92), was integrated with the NSGA-II algorithm to identify compromise solutions between productivity and wear. The results provide a Pareto Frontier with practical recommendations for different productive strategies. The approach proved to be effective in modeling and optimizing machining processes, with potential for direct application in the metalworking industry.