Information properties of vibroacoustic emission in diagnostic systems for cutting tool wear

Introduction. This paper is devoted to the development of a methodology for diagnosing cutting tool wear based on the analysis of vibroacoustic emission signals. Two tasks are addressed. Firstly, the information feature space related to wear is constructed. Secondly, within this space, decision rules are defined that allow division into separate clusters according to wear levels. Since the construction of the information feature space (IFS) methods is of primary importance in these procedures, the purpose of this work is to determine the regularities of changes in the frequency characteristics of the dynamic cutting system caused by wear development and to construct, on this basis, a rational information space for diagnosing tool wear. Method and methodology. The study is based on mathematical modeling results of a perturbed dynamic cutting system to determine the information feature space representing tool wear. Methods for determining the parameters of information signal parameters (ISPs) are proposed, which provide high sensitivity to wear changes. All ISP parameters should be dimensionless and zeroed at zero wear. They must satisfy additional requirements, including noise immunity conditions. Results and discussion. The paper presents results of constructing ISP parameters for vibroacoustic emission analysis in two frequency ranges. In the low-frequency range, limited by the first natural frequencies of interacting subsystems (up to 1.0–1.5 kHz), vibration response parameters (VRP) are determined based on vibration sequences obtained analytically under power perturbations modeled as “white” noise. In the high-frequency range (above 2.0 kHz), information models based on random pulse sequences are proposed. It is shown that the applicability of a particular information feature depends on the conditions. Thus, the developed methodology, mathematical simulation, and digital and field experiments enabled the formation of a rational information space for wear diagnostics, in which known recognition methods can be used to construct decision rules for classifying information according to wear levels.