TOOL CONDITION MONITORING FOR TURNING BY ACOUSTIC EMISSION

Abstract

This project investigates the possibility and effectiveness of using acoustic emission (AE) technique to indirectly monitor the tool condition and predict tool failure in turning. The aim is to determine the potential of using acoustic emission for in-process tool condition monitoring in an unmanned factory. Experimental work has been conducted using different types of tool inserts and workpiece materials under various cutting parameters. The signal from the AE sensor is first amplified and :filtered, then sampled and digitized, and finally stored and analyzed in a computer. Software programs have been developed to continuously acquire the AE data, perform signal processing, and identify suitable parameters for the prediction of tool failure. Findings show that there is a good correlation between the signal of acoustic emission and the state of the cutting tool during turning operations. From the acquired data, appropriate AE signal parameters are analyzed in the time domain as well as the frequency domain. A general consistent trend is observed when the AE signal is presented in the RMS voltage and in the band power, especially in the higher frequency range. The RMS voltage and the band power of the AE signal are generally characterized by a steep increase at the initial tool "run-in" stage, a slow increase as the tool wear progresses, and a sharp increase during the accelerated wear stage till tool failure. The amplitude and spectral distribution of the AE signal are also analyzed. As tool wear progresses, the amplitude distribution of the AE signal shifts to higher amplitude area over a broader frequency band. From the power spectral density plots, increasing AE activities are observed in the high frequency range of 300-600 KHz, especially when the tool approaches failure. The trends are found to be similar in both failure cases of excessive wear and sudden fracture. The results are verified by the tool images captured during turning operations. It is deduced that the AE signal increases when the friction between the tool and the workpiece becomes larger as a result of progressive tool wear. As the cutting tool reaches failure, fracture or excessive wear generates the burst-type AE signal of high amplitude and causes the band power to increase significantly in the high frequency range (300-600 KHz). Therefore, the onset of tool failure can be predicted by the AE signal characteristics before the tool fails. In this study, it is established that the use of the acoustic emission technique to monitor the tool condition is feasible and effective. Furthermore, the potential of acoustic emission can be enhanced by incorporating suitable high-performance hardware and by carrying out further investigation in signal processing and sensor fusion techniques.