Tool Condition Monitoring based on sound and vibration analysis and wavelet packet decomposition

Abstract

Tool Condition Monitoring (TCM) is a vital demand of advanced manufacturing in order to develop automated unmanned production. Continuing the machining operation with a worn or damaged tool will result in damages to the workpiece. This problem becomes more important in supplementary machining processes like drilling which the workpiece usually has passed a lot of machining processes and any damage to workpiece at this stage results in high production losses. In this research features of sound pressure and vibration signals in drilling process are recorded and analyzed in order to detect tool wear. Signal statistical features are extracted in time domain, and the features trends as the tool becomes worn are extracted. Frequency spectrum of signals is calculated and Wavelet Packet Decomposition (WPD) is applied to focus on specific frequency bands. In this research capability of both sound and vibration signals for drill wear detection are shown and the most informative features of the signals for wear detection are evaluated and introduced. The results showed that the wavelet packets features make a better contrast between the sharp and the worn tool compared to the primary time domain signal.