Data-driven modeling and optimization of tribological performance of grey cast iron under dry and lubricated sliding

Grey cast iron is extensively employed in tribological components owing to its excellent castability, and cost-effectiveness. However, its sliding wear behavior under varying sliding conditions requires systematic investigation and reliable predictive modeling. In the present study, the tribological performance of grey cast iron was systematically evaluated through experimental analysis under dry and oil-lubricated conditions. A total of 32 experimental datasets were generated under varying applied loads and sliding speeds to evaluate the wear behavior and frictional response of the test material. Microstructural characterization revealed randomly distributed graphite flakes embedded within a predominantly pearlitic matrix. The experimental results revealed significantly higher wear rate and coefficient of friction (COF) under dry sliding, whereas oil lubrication substantially reduced friction and material loss. A two-level stacking ensemble learning framework was developed to predict tribological responses. The stacking model demonstrated excellent predictive capability with R 2 values of 0.99 for wear rate and 0.97 for COF. Subsequently, single-objective optimization using the Fruit Fly Optimization Algorithm (FOA) identified optimal operating parameters under oil-lubricated conditions, achieving a minimum wear rate of 1.633 × 10 −14 m 3 /m and a minimum COF of 0.1077. Experimental validation confirmed a close agreement between the predicted and measured responses. The developed stacking–FOA framework provides an effective data-driven approach for prediction and optimization of tribological performance of grey cast iron.