Determination of stability of the cutting process under dynamic loading conditions

Abstract

An approach to quantitatively determining the stability of the cutting process under dynamic loading conditions is demonstrated. With this method, it is proposed to evaluate the time series of cutting force values and determine the stability index as a function of the prediction accuracy, which is determined by the root mean square error of deviation (RSME) of the predicted and experimental values of the cutting force signal. The prediction of the dependence is carried out using machine learning based on the XGB method (extreme gradient acceleration). The practical significance of the developed method lies in the ability to provide a quantitative assessment of the stability of the cutting process regardless of the state of the cutting edge of the tool, which can be characterized by both uniform wear and the presence of defects (microcracks, chips) and macrofracture. Experimental values of the cutting signal strength were obtained when processing a workpiece made of hardened steel with longitudinal grooves to simulate a pulsed dynamic load on the cutting tool with PcBN «Borsinit». The cutting force values were recorded using a UDM-600 dynamometer and an ADA-1406 ADC. It was found that the RSME values in the range of applied processing conditions (v = 120–210 m/min; S = 0.10–0.19 mm/rev; t = 0.1–0.2 mm) were within 40–60 N. The optimal parameters of cutting modes that ensure stable cutting under dynamic conditions correspond to the following conditions: longitudinal tool feed S = 0.10–0.12 mm/rev; cutting depth t = 0.125–0.175 mm; cutting speed v = 180–200 m/min. at a value of RSME < 56. It is also shown that the use of a tool with a cutting edge inclination angle λ = 40–50° practically eliminates cutter breakage and increases the stability of the cutting process.

Keywords

• cutting tool
• cutting force
• intermittent machining
• process stability
• machine learning
• prediction
• tool failure
• wear