Multi-objective optimization of SUS430C steel turning process using hybrid machine learning and evolutionary algorithm approach

• Combines methods Extreme Gradient Boosting (XGBoost) and Non-dominated Sorting Genetic Algorithm-II (NSGA-II) to optimize three important machining objectives: surface roughness (Ra), material removal rate (MRR) and tool wear (Vb). • These models served as surrogates for fitness evaluation in NSGA-II, enabling efficient multi-objective optimization. The results yielded 16 Pareto-optimal solutions that balance the trade-offs among Ra, MRR, and Vb. Notably, the study highlights the importance of feed rate (fz) and depth of cut (ap) in influencing R a and Vb, while cutting speed (Vc) significantly impacts MRR. • The optimal parameters for achieving the lowest R a were Vc=183.01 m/min, fz=0.08 mm/rev, and ap=1 .22mm. This study focuses on the turning process of SUS430C stainless steel, a ferritic stainless steel known for its excellent corrosion resistance and moderate mechanical properties, commonly used in automotive and kitchen applications, a material widely used in industrial applications but challenging to machine due to its hardness and work-hardening characteristics. A hybrid approach combining Extreme Gradient Boosting (XGBoost) and Non-dominated Sorting Genetic Algorithm-II (NSGA-II) was employed to optimize three critical machining objectives: surface roughness (R a ), material removal rate (MRR), and tool wear (V b ). The predictive capability of XGBoost models was validated using experimental data from a Box-Behnken design, achieving high accuracy with R 2 values exceeding 0.93 across all performance metrics. These models served as surrogates for fitness evaluation in NSGA-II, enabling efficient multi-objective optimization. The results yielded 16 Pareto-optimal solutions that balance the trade-offs among R a , MRR, and V b . Notably, the study highlights the importance of feed rate (f z ) and depth of cut (a p ) in influencing R a and V b , while cutting speed (V c ) significantly impacts MRR. The optimization framework provided practical insights into machining parameter selection, with the lowest R a of 0.85 µm achieved at V c =183.01 m/min, f z =0.08 mm/rev, and a p =1.22 mm. The findings underscore the effectiveness of the hybrid XGBoost-NSGA-II approach in solving complex manufacturing optimization problems and serve as a foundation for future applications in sustainable and efficient machining practices.