PREDICTION THE MAXIMUM HEIGHT OF ROUGHNESS PROFILE AS A FUNCTION OF TOOL WEAR USING ARTIFICIAL NEURAL NETWORKS
Radoslav Vučurevi?ć,
Zdravko Krivokapić,
Brankica Čomić
Abstract: By training artificial neural networks based on data obtained through experiment conducted using the Taguchi design in drilling process, this study developed models for predicting the maximum height of roughness profile as a function of tool wear. The development of models with multiple inputs and a single output was performed using Backpropagation artificial neural networks with one and two hidden layers, employing sigmoid transfer functions in the hidden layers and a linear transfer function in the output layer. During the models development, the input parameters of the drilling process and the tool wear values before tool blunting were used as network inputs, while the maximum height of roughness profile was used as the network output. A comparative analysis of models errors, calculated based on experimental values and predicted maximum height of roughness profile values for given input parameters and wear at the moment of tool blunting, led to the conclusion that the best results were obtained using the artificial neural network model with two hidden layers, employing sigmoid transfer functions in the hidden layers and a linear transfer function in the output layer.
Keywords: Prediction, Roughness, Tool wear, Artificial neural networks
DOI: 10.24874/IJQR20.02-08
Recieved: 10.09.2025. Accepted: 02.03.2026 UDC:
Reads: 123 
