Unsupervised Machining Recognition from a Vibration Signal

Abstract Objectives In unattended machining, monitoring the cutting processes holds significant importance. A crucial aspect to consider is whether chip formation occurs at a specific moment or not. Acquiring this knowledge can enhance production reliability and minimize losses. Methods To address this, an algorithm is proposed for detecting chip formation using vibration signals obtained from a CNC machine. The algorithm relies on the spindle RPM and vibration signal to identify the occurrence of machining cycles (chip formations). The evaluation process involved comparing the results of the algorithm to those obtained by a human manually selecting the machining parts from the same signal. Results Its effectiveness was evaluated across various feedrates, demonstrating its applicability under different conditions. Conclusions This method is able to detect the real machining more accurate than the use of G-code-based method.