Tool wear model based on least squares support vector machines and Kalman filter

Abstract

In this study, the least squares support vector machines (LS-SVM) and Kalman filter (KF) technique are used to establish the tool wear estimation model. Tool wear prediction model, based on LS-SVM, is given to describe the mapping relationship between input–output factors. The cutting conditions (feed rate, cutting speed, and depth of cut), cutting time, and wear position constitute the input factors and tool wear is the output parameter of the model. In order to improve the accuracy of the LS-SVM results, the KF technique is used to update the tool wear estimated results of LS-SVM-based model, which is called the LS-KF model, according to the measured tool wear values. Experiment work is performed on machining center for cemented carbide ball-end cutter cutting stainless steel. Those two models (LS-SVM model and LS-KF model) are applied to the actual milling machining to verify their performance. Results show that predicted tool wear based on the proposed LS-KF model has more precision than that of LS-SVM model.