Monitoring and analysis of ultra-precision machining processes using acoustic emission

Effective in-process monitoring techniques are essential to achieving high precision and productivity in ultra-precision machining. The potential of acoustic emission (AE) for monitoring the fine process parameters in micro-machining was investigated and novel AE signal representation and feature extraction techniques were developed.; Spectral analysis of acoustic emission signals from micro-cutting experiments showed that the specific AE RMS increases sharply as the uncut chip thickness approaches the tool edge radius, indicating redundant energy from plowing and sliding rather than chip formation. Experimental results also showed that acoustic emission is very sensitive to the process parameters in ultra-precision metal cutting, and is directly related to the material removal mechanisms. Any quantitative model of AE energy in micro-machining should account for the "edge-radius effect". A round-edge tool force model was investigated for deriving an AE energy model in ultra-precision metal cutting and verified with cutting force data.; An analysis of the link between wavelet analysis and the inherent Green's function transfer characteristics of AE wave propagation was done. Characteristic waveforms for brittle and ductile source events were summarized and used as the basis for identifying material removal mechanisms. By incorporating Green's function solutions with wavelet analysis, a representation scheme for AE signals was developed. Numerical simulations demonstrated the efficiency of the proposed scheme in extracting key features correlated to AE source events. The wavelet packet analysis is highly efficient in locating the onset of individual bursts, even with overlapping bursts.; The proposed signal representation scheme was then applied to the identification of cutting states for metal micro-cutting and the brittle-ductile transition for brittle materials. It was found that the time-frequency representations from wavelet analysis contain distinct patterns for identifying the material removal mechanisms, including burr formation. Parameters such as burst rate and entropy were proposed as quantifiers of the relative "brittleness" of the machining process.; Strategies for integrating wavelet analysis and neural networks were discussed. An intelligent in-process monitoring system for controlling cutting states and tool condition in micro-machining was developed. Wavelet analysis extracts the most descriptive features for neural network processing and enhances the performance of the networks.