Precision turning of CK45 steel for calibration devices requires stringent control of surface characteristics, geometrical accuracy, and tool wear. This paper introduces a hybrid intelligent modeling and optimization framework that combines Response Surface Methodology (RSM) with Artificial Neural Networks (ANN) to simultaneously forecast machining efficiency and correlate it with microstructural evolution. Experiments were carried out via cubic boron nitride (CBN) tools under a central composite design, considering spindle speed (N), feed rate (F), depth of cut, (D) and tool nose radius (R) as key process parameters. The effects of these variables on material removal rate (MRR), tool wear rate (TWR), surface roughness parameters (Ra and Rmax), roundness error (OR), and hardness (H) were systematically analyzed. While RSM regression models provided initial predictive capability, their integration with ANN significantly enhanced accuracy, yielding prediction errors below 7% for all responses. Multi-objective optimization identified machining conditions that minimized tool wear (1.23 × 10⁻⁵ g/min) while improving surface quality and dimensional precision. Microstructural analysis further revealed the formation of refined and homogeneous dendritic structures with characteristic sizes ranging from 10.86 to 17.44 μm under optimized conditions. The proposed hybrid intelligent RSM–ANN framework offers a robust and reliable approach for precision turning optimization and enables new insights into the linkage between machining parameters and microstructural characteristics of CK45 steel.
Farouk et al. (Thu,) studied this question.