Tool life prediction for sustainable manufacturing

Abstract

Prediction of tool wear is essential to maintaining the quality and integrity of machined parts and minimizing material waste, for sustainable manufacturing. Past research has investigated deterministic models such as the Taylor tool life model and its variations for tool wear prediction. Due to the inherent stochastic nature of tool wear and varying operating conditions, the accuracy of such deterministic methods has shown to be limited. This paper presents a stochastic approach to tool wear prediction, based on the particle filter. The technique integrates physics-based tool wear model with measured data to establish a framework, by iteratively updating the tool wear model with force and vibration data measured during the machining process, following the Bayesian updating scheme. Effectiveness of the developed method is demonstrated through tool wear experiments using a ball nose tungsten carbide cutter in a CNC milling machine.