Prediction of residual stress in wire electrical discharge machining of aluminum carbon silicide metal matrix composite using an artificial neural network approach

Machining of aluminum–Silicon carbide (Al/SiC) metal matrix composites (MMCs) using conventional machining methods is difficult due to the presence of hard and abrasive SiC reinforcement particles, which cause tool wear and poor surface integrity. Consequently, non-conventional machining methods are preferred for processing such composites. Among these, Wire Electrical Discharge Machining (WEDM) is widely practiced due to its ability to machine complex geometries with high precision. However, in WEDM, effective control of residual stress is critical to ensure the structural integrity and long service life of Al/SiC MMC components. The aim of this project is to develop an Artificial Neural Network (ANN) model predicting the residual stress value in WEDM of Al/SiC MMC, considering five input parameters: Pulse on Time, Pulse off Time, Peak Current, Servo Voltage, and Wire Tension. Coefficient of determination R 2 values of 0.99996 for training, 0.99999 for testing, and 0.99998 for validation show that the ANN model successfully captures the relationship between WEDM parameters and residual stress and lessens the need for lengthy experiments. The project concludes that ANN-based prediction is a reliable tool for optimizing WEDM processes and improving the mechanical performance of Al/SiC MMC components.