Experimental Investigation and Machine Learning Modeling of Electrical Discharge Machining Characteristics of AZ31/B₄C/GNP_s Hybrid Composites

Magnesium alloys, like AZ31, possess a desirable low weight and high specific strength, which make them favorable for aerospace and auto applications, yet their difficulty to machine limits their broader implementation for the industry. Electrical discharge machining (EDM) is an effective technology for machining difficult-to-machine materials, particularly when the materials are reinforced with ceramic and graphene-based fillers. This study examines the impact of reinforcement percentage (R) and different electrical discharge machining (EDM) parameters such as current (I), pulse on time (T_on) and pulse off time (T_off) on the material removal rate (MRR) and surface roughness (SR) of AZ31/B₄C/GNPs composites. The combined reinforcement range varies from 2 wt.% to 4 wt.%. The Taguchi design (L27) is utilized to conduct the experiments in this study. ANOVA of the experimental data indicated that current (I) significantly affects MRR and SR, exhibiting the greatest contribution of 44.93% and 51.39% on MRR and SR, respectively, among the variables analyzed. The surface integrity properties of EDMed surfaces are examined using SEM under both higher and lower material removal rate settings. Diverse machine learning techniques, including linear regression (LR), polynomial regression (PR), Random Forest (RF), and Gradient Boost Regression (GBR), are employed to construct an efficient predictive model for outcome estimation. The built models are trained and evaluated using 80% and 20% of the total data points, respectively. Statistical measures (MSE, RMSE, and R²) are utilized to evaluate the performance of the models. Among all the developed models, GBR exhibited superior performance in predicting MRR and SR, achieving high accuracy (exceeding 92%) and lower error rates compared to the other models evaluated in this work. This work demonstrated the synergy between techniques in optimizing EDM performance for hybrid composites using a statistical design and machine learning strategies that will facilitate greater use of hybrid composites in high-precision engineering applications and advanced manufacturing sectors.