Statistical- and data-driven optimization of Al ₂ O ₃ nanoparticle-assisted electrochemical machining of Inconel 718 using machine learning models

Machining nickel-based superalloys such as Inconel 718 remains a challenging task due to their high hardness, strength and thermal resistance, which often result in low material removal rates (MMRs) and suboptimal surface quality using conventional methods. This study presents a statistical and data-driven framework to enhance electrochemical machining (ECM) performance by incorporating aluminum oxide (Al2O3) nanoparticles in the electrolyte, coupled with machine learning (ML)-based optimization. An extensive experimental dataset consisting of 27 runs was generated by systematically varying applied voltage, feed rate (FR) and electrolyte discharge rate (EDR). Random forest (RF) regressor models were developed to predict surface roughness (SR) and MRR as functions of these parameters, achieving high predictive accuracy with [Formula: see text] values of 0.9701 for SR and 0.9831 for MRR, and low mean absolute errors (MEAs) of 0.0506[Formula: see text] [Formula: see text]m and 1.2198[Formula: see text]m[Formula: see text]/min, respectively. Multi-objective optimization identified the optimal parameter combination — voltage: 18.0[Formula: see text]V, FR: 0.5[Formula: see text]mm/min and EDR: 12.0[Formula: see text]l/min — yielding a minimum SR of 1.038[Formula: see text] [Formula: see text]m Ra and a maximum MRR of 117.9[Formula: see text]mm 3 /min. The strong correlation between predicted and experimental results validates the robustness of the approach. This work demonstrates that integrating nanoparticle-assisted ECM with ML provides a physically interpretable and efficient methodology for precision machining of difficult-to-cut materials. The framework offers a transferable strategy for process optimization in advanced manufacturing, highlighting the interplay of statistical modeling, material properties and process dynamics.