Comparing the reliability of classical statistics and data mining techniques in unit energy prediction in circular stone cutting

Abstract

Energy efficiency is one of the critical parameters affecting production in the natural stone sector, as it is in every industrial sector. High energy consumption negatively affects production costs, especially in stone cutting and surface treatments. Nowadays, it is crucial to predetermine energy consumption with reliable predictive techniques to produce with the lowest energy possible and sustain sectorial competition. This study conducted stone cutting tests with a computer-assisted circular cutting machine at different peripheral speeds (PSs) and advance rates (ARs). Unit energy (UE) consumptions were measured in stone cuttings.UE was evaluated regarding the circular stone cutting machine's (CSCM's) operating parameters, some stone characteristics, vibration amplitude (VA), and sound level (SL) measured during cutting. Classical statistics (CS) and data mining (DM) techniques were used to predict UE. 287 and 24 cutting data sets were selected as training and testing data for CS and DM techniques, respectively. These techniques were also compared and provided more significant and reliable results of DM techniques than CS. DM techniques predicted the UE with high correlation coefficients obtained in the range of R 2 =0.963 and 0.973. DM models for UE prediction before stone cutting have been introduced for stone processing researchers and those interested.