Machined quality prediction and optimization for micro-EDM drilling of semi-conductive SiC wafer

Semi-conductive SiC (semi-SiC) wafer is the third generation of semiconductor materials. However, its high hardness and brittleness make it extremely difficult to machine 3-dimensional complex shapes by conventional machining methods. Electric discharge machining (EDM) is potentially an alternative method for machining semi-SiC wafer. In this study, micro-hole drilling by EDM on semi-SiC wafer was investigated using an assisting electrode technique. A deep neural network (DNN) model was developed for predicting the machined quality. The adjustable machining parameters considered were pulse-on time, pulse-off time, peak current, and working time. The machined quality indices included the inlet and outlet diameters of the EDMed hole. The importance and effects of machining parameters on the machined quality were analyzed using Random Forest Method and Response Surface Method, respectively. Particle Swarm Optimization (PSO) algorithm was employed to find the optimal DNN architecture. The developed DNN model was effective in predicting machined quality, as demonstrated by low mean absolute percentage error (MAPE), low mean squared error (MSE), low root mean squared error (RMSE), and high coefficient of determination (R²) in the training and testing processes. The developed DNN model coupled with the PSO was used to find the optimal machining parameters for generating the best combination of machined quality. Further validation experiments were conducted and the results verified that the DNN model and PSO were capable of predicting and optimizing the machined quality for EDM of semi-SiC wafer.