Effective RUL prediction of the milling tool based on Shannon entropy feature selection method and squared exponential-Gaussian process regression model

Online condition monitoring and remaining useful life (RUL) prediction play an essential role for cutting tools in the machining process as they directly affect the product’s quality and reliability. This work proposes an efficient and reliable method for RUL prediction of the cutting tools using decision-making models such as the squared exponential Gaussian process regression (SE-GPR) method. The work also defines the optimum feature selection approach based on Shannon entropy. Firstly, the vibration and force sensor features are extracted, and their Shannon entropy values are calculated. Then, the features with minimum Shannon entropy values are selected that carry the maximum information about the cutting tool state and effectively reflect the cutting tool’s degradation patterns. Finally, the SE-GPR model is utilized to examine the RUL prediction capability of the selected features. The proposed Shannon entropy-based feature selection method effectively reduces the number of features and improves computational and prediction accuracy compared to other feature selection methods, i.e., principal component analysis, Pearson correlation coefficient, and suitability index (monotonicity, trendability, and robustness). The prediction accuracy has been enhanced using selected features and the SE-GPR model, showing its superiority for RUL prediction compared to other approaches such as neural networks and SVM.