Machining process characterization and intelligent tool condition monitoring using acoustic emission signal analysis

Smart sensor systems are expected to play an important role in the development of successful untended machining centers. The objective of this thesis is to characterize acoustic emission generated during a machining operation and to develop an acoustic emission-based intelligent sensor system for tool condition monitoring.;Acoustic emission (AE) generated during orthogonal cutting experiments was analyzed. Experimental results showed that the power in the AE signal was proportional to the square of the cutting velocity. Based on this observation, it was suggested that damping of dislocation motion during its flight period between obstacles was responsible for generation of AE at the high strain rates encountered in metal cutting. The power content of the AE signal was also found to depend on the chip-tool contact length, with high signal power being observed at very low values of the contact length. This was attributed to high strain rates associated with thinning of the shear zone. The spectral characteristics of the AE generated during orthogonal machining were also studied.;Tool wear tests were conducted on a lathe, using conventional carbide inserts, in order to determine the sensitivity of AE energy and spectral characteristics to tool wear. It was found that although AE was sensitive to tool wear, it was also fairly sensitive to the effects of process variables and noise, so that tool wear monitoring using only AE information is a difficult task. A pattern recognition system using AE and cutting force information along with information on process variables (such as the feed rate and cutting velocity) was developed in order to associate incoming sensory patterns with the tool wear status. A further development in this work involved application of feedforward neural networks for learning and pattern association tasks. The results of this work show that neural networks offer a powerful method for implementing intelligent, multiple sensor systems for real time process monitoring applications.;The effect of sensor placement on the detected AE was studied for a turning operation. It was found that although the AE signal was extremely noisy due to rotation of the bearing elements, it exhibited sensitivity to tool wear and tool breakage, showing that AE signals transmitted through bearing surfaces retain sufficient sensitivity to the mechanics of the cutting process. This is of great significance in many turning, milling and grinding operations in which the outer surface of a rotary bearing may be the only practical site for sensor location. (Abstract shortened with permission of author.).