Machine Learning Based Predictive Modeling of Machining Performance of Ti-6Al-4V under MQL Lubricating Conditions

Abstract

This study evaluates the performance of machined Ti-6AI-4V alloy under Minimum Quantity lubrication (MQL) using machine learning models to support sustainable and efficient milling. Experiments were conducted by varying cutting speed, feed rate, and their effects on surface roughness (Ra), material removal rate (MRR), and tensile strength (Ts) have been recorded. Using the experimental dataset, predictive models were developed using Random Forest (RF) and Artificial Neural Networks (ANN) to estimate machining responses. RF has employed a collection of decision trees to boost prediction stability, whereas ANN captured nonlinear. Models’ performance was evaluated using R2, root mean square (RMSE) and mean absolute error (MAE) metrics. The outcomes showed that both models predicted machining responses under MQL conditions efficiently; however, the ANN model demonstrated superior accuracy, particularly for MRR and Ts. Statistical evaluation confirmed that the ANN achieved the highest predictive accuracy particularly for Ts with R2 = 0.9404, RMSE = 0.0737 and MAE = 0.038 while also demonstrating reliable prediction for Ra and MRR. The RF model showed lower accuracy particularly for MRR yet maintained stable regression behaviour which indicated its ability to capture non-linear patterns despite higher prediction errors. Overall, the developed models offer a reliable data-driven framework for optimizing machining parameters and improving sustainability.