A comparison of selected acoustic signal analysis techniques to evaluate wood surface roughness produced during routing

This study examines the applicability of several signal analyzing methods applied to audio signals to establish relationships between audio signals generated during machining and the surface roughness produced. One-third octave passband filters were also used to condition audio signals acquired during routing of Japanese beech (Fagus crenata Blume) at different machining conditions. The one-third octave analysis was essential for gaining satisfactory results. Three different analyzing strategies were investigated: the RMS averaging method and the count rate method combined with static and dynamic thresholds. The count rate method at constant threshold could not be applied to the sound intensity signal due to its substantial range in magnitude during machining at different states, although the dynamic threshold strategy can be applied to both the sound intensity and the sound pressure signals, and adjusts flexibly to the variable signal generated. Linear regression of data obtained during each of the analyzing strategies shows that the RMS averaging method is most suitable for analyzing audio signals, as shown by its highest correlation coefficient with surface roughness produced. The results suggest that measurement of sound intensity is a promising method for monitoring surface roughness and determining optimum feed rates.