A Novel Hybrid RSM-ANN Model for Surface Roughness Prediction in Turning of Al 6061 Alloy

Surface roughness prediction based solely on cutting parameters provides a quantified value regardless of tool condition, making it suitable for initial parameter selection. However, to achieve accurate surface roughness prediction, the current tool condition must be incorporated into the model. This can be accomplished by utilizing vibration or cutting force signals. In this study, we develop a hybrid response surface methodology-artificial neural network (RSM-ANN) model for predicting surface roughness by combining cutting parameters and vibration data. Four different RSM models were developed, and the best-performing model was selected as input for the hybrid RSM-ANN model. A comparison was made between the hybrid model, a basic ANN model with four inputs (three cutting parameters and one mean vibration in the [Formula: see text]-direction), and other ANN models with all possible combinations of these four variables. The hybrid model demonstrated the highest accuracy with the mean square error of 0.00332 with the highest coefficient of regression value of 0.99802 when compared to the other models.