Co-authors​ :

 Xiaoyu WANG (CHINA), Yan CHEN , Jihong HE , Liangtao HU (CHINA) 

Carbon Fiber Reinforced Plastics (CFRP) are widely used in aviation and aerospace due to their exceptional physical properties. However, the high strength and hardness of carbon fibers also lead to rapid tool wear, thus resulting in significant surface damage and delamination. Therefore, online monitoring of tool wear is imperative to ensure that any processing anomalies and defects are maintained within acceptable limits. Edge trimming experiments were conducted on CFRP with router tools, where a suite of signals, including acoustic emission, cutting force, acceleration, and temperature was collected during the cutting process. The results indicated that the signals were modulated by the dynamic rigidity shifts. Wavelet transform and Hilbert-Huang transform (HHT) were used to analyze the periodic characteristics of the signal for distinguishing the authentic cutting processes from the mechanical chatter. The degree of correlation between signal characteristics and tool wear was evaluated using the Pearson coefficient. Investigative efforts into the time-domain signal characteristics showed that certain parameters, such as peak amplitude, average value, and root mean square value, displayed relatively low and stable readings during the initial stages of tool wear, and evolved to larger and unstable wave amplitudes as the tool wore further. The Pearson coefficients of cutting force and tool flank wear were 0.94, 0.96 and 0.96, higher than that of acceleration 0.87, 0.76 and 0.91, which increased by 8.05%, 26.32% and 5.49%. Relative to acoustic emission, the increases are 54.10%, 12.94% and -1.03%. The Pearson coefficient of temperature was 0.95, which was slightly higher than cutting force, but other characteristics were lower than 0.5. In dimensionless characteristics, the kurtosis and margin factor of the cutting force reached 0.88 and 0.73, showing an increase of 2.94% and 12.72% compared to the acceleration, and 35.38% and 6.48% compared to acoustic emission. The trend in the parameters of the frequency-domain signal was investigated, including center of mass frequency，standard deviation of frequency and the RMS frequency. The Pearson coefficient of cutting force were -0.64, -0.76, and 0.84, respectively, with an increase of -3.15%, 2.15%, and 7.45% compared to the acceleration, and an increase of -10.26%, 6.95%, and -0.97% compared to acoustic emission. The above results indicated that in terms of sensitivity to tool wear, the first was cutting force, followed by acoustic emission, acceleration, and temperature. The aim of this paper is to demonstrate the potential for online monitoring of the cutting process through different signals, with assessing the sensitivity of the relationship between the signals and tool wear through detecting the machining accuracy of the material surface, with suggestions of evaluation and monitoring methods of tool wear state using mainly cutting force signals and supplemented by other signals.