Multi-categories tool wear classification in micro-milling

Abstract

Micro-milling is a precision/ultra-precision milling technology where the cutting parameters are in micro scale. Tool condition monitoring (TCM) in micro milling poses new challenges compared to conventional machining, due to the high tool wear rate and high precision requirement associated with the dimensions to be produced at micro-level. First, the noise component in the signal for monitoring micro-milling is usually very high and difficult to separate. Second, the micro-milling is highly non-stationary and involves spike signals, and this brings difficulty for the tool state estimation. Third, multi-category classification is required due to its high precision product requirement. In this thesis, a noise-robust multi-category classification approach for tool flank wear state identification in micro-milling is developed for this purpose. Continuous Hidden Markov models (HMMs) are adapted for modeling of the tool wear process and estimation the tool wear state based on the cutting force features. In the framework of HMMs, the thesis develops a tool for signal denoising first when the desired signal is contaminated with non-Gaussian noise and the noise spectrum distribute widely in the frequency domain; and then develops a noise-robust multi-category tool wear state classification system for micro-milling. Experimental studies on the tool state estimation in the micro-milling of pure copper and steel illustrate the effectiveness and potential of these approaches.