MONITORING AND ANALYSIS OF THE MACHINING PROCESS USING ACOUSTIC EMISSION SENSING TECHNIQUES (SENSOR, CUTTING)

This study emphasizes sensor development for the automated machining process. A thorough review was made for all the major sensors in the application of machining process monitoring. Acoustic emission (AE) shows very high potential in this application due to its close association with the material structure changes. It is the goal of this study to make AE a useful sensor for machining process monitoring through real applications with the theoretical support.; The AE technique was first applied to monitor the diamond-turning process in order to guarantee a high precision operation. A normalization process was adopted to reduce the unexpected influence on the measured signal. A stable cutting can generate a constant value of the root mean square (RMS) of AE signal. The data set of RMS signal obtained from same cutting tool and workpiece material were fitted by both second order polynomial and power function with cutting parameters, like cutting speed, feed rate, and depth of cut, as variables. The derived functions can be used to estimate the RMS level of AE for the diamond turning within a specified range of each parameter. The tests showed that the estimated values were generally within 5% accuracy of the measured values.; The AE technique was also applied to another important area of unmanned machining--chip form detection. The AE signal was proved to contain the information of chip breaking and the RMS signal has different characteristics for different chip forms. A methodology was developed to use the multiplication of the average RMS level times a threshold multiplier as the threshold to measure the event rate of RMS signal. A median event rate indicates the discontinuous-chip cutting, and a low event rate indicates continuous-chip cutting.; In order to resolve the deficiency in dealing with the special chips above and the ambiguity around the transition from continuous to discontinuous chip, an enhanced signal processing capability was needed. The linear discriminant function, which is a pattern recognition technique, was adopted for chip form classification. The pattern recognition technique worked very well in chip form identification. (Abstract shortened with permission of author.)...