A technique for flank wear monitoring in milling machines based on the novel tool wear indicator

Abstract

ABSTRACT This research investigates an innovative method for monitoring flank wear by analyzing these signal variations. By dissecting the signal bands into multiple frequency bands using Wavelet Packet Transform, a precise examination of the signal contents is enabled. Using the signal power spectrum to Shannon entropy (SPSSE) node power spectrum analysis metric, the specific frequency band that exhibits a stable rising trend in the proposed metric correlated with flank wear is identified. This frequency band serves as a basis for constructing a tool wear indicator that utilizes Mean Square Error (MSE) comparisons of a healthy baseline signal sample with the sample under test. The calculated indicator closely correlates with the actual measured levels of flank wear, allowing for effective evaluation of the tool condition and differentiation between the healthy tool and the tool exhibiting any levels of flank wear. The performance of the proposed method is evaluated using the publicly available benchmark NASA Ames Milling tool dataset. This innovative approach holds promise for real-time assessment of machining processes, facilitating timely intervention based on the tool wear estimation.