선삭 가공 중 발생하는 음향 신호 기반 장비･소재･공정 특성 분석 및 랜덤 포레스트를 이용한 식별 가능성 연구

초록

This study investigated acoustic signatures in turning hollow cylindrical workpieces, focusing on machine-, workpiece-, and process-induced frequencies and chatter. ANSYS modal analysis showed bending-mode frequencies decreased by up to 28% as workpiece length increased, while oval modes remained nearly constant. Turning tests on two CNC lathes identified machinespecific frequencies at 578 and 904 Hz, independent of workpiece geometry. Workpiece-induced frequencies matched modal predictions: the 113 mm workpiece exhibited bending and oval modes near 2.7 and 5.8 kHz, respectively, while the 93 mm workpiece showed an oval mode near 6.2 kHz. Outside diameter (OD) roughing, OD finishing, and insider diameter finishing were spectrally distinguishable. Chatter occurred at natural frequencies and harmonics but was reduced by decreasing the tool nose radius from 0.8 to 0.4 mm. A random forest classifier identified machine, workpiece, and process conditions with accuracies of 96%, 79%, and 63%, respectively.

키워드