Acoustic signal-based tool condition monitoring in belt grinding of nickel-based superalloys using RF classifier and MLR algorithm

With outstanding material removal ability and high finish quality, robotic belt grinding has great advantages in processing difficult-to-machine materials like nickel-based superalloys. Tool wear is a severe problem in such grinding processes; thus, detection of tool wear is critical to precision finishing of a surface profile. This work proposes a novel acoustic signal-based detection method that combines a random forest (RF) classifier and a multiple linear regression (MLR) model to detect different wear periods and evaluate the remaining grinding ability for robotic belt grinding of nickel-based superalloys. The correlation between grinding sound and belt conditions is established through experimental studies and signal analysis. Through mapping the acoustic features of grinding sound and conditions of grinding belts, the RF classifier and the MLR model are trained and applied in prediction of grinding belt conditions. The total prediction accuracy of RF classifier for distinguishing different wear periods is over 94%, and the mean absolute percentage error of MLR model for evaluating the grinding ability in accelerated wear period is less than 9%. The online detection method can be used as a basis for adaptive control of grinding parameters to achieve precision profile finishing.