Abstract
This study presents a comprehensive analysis of defect detection in the manufacturing process of solid carbide milling tools. The creep-feed flute grinding technique was used to fabricate a milling tool, with cutting force signals recorded and examined using recurrence analysis and conventional statistical methods. The analysis identified four distinct dynamic fluctuations (cutting force amplitude jumps), which showed a direct correlation with the formation of microcracks on the flute surface. These jumps exhibited varying levels of reduction, ranging from 5% to 22% in amplitude. A detailed investigation, including recurrence plots and recurrence quantification analysis (RQA) with a moving-window approach, revealed that several recurrence indicators, such as the recurrence rate (RR), determinism (DET), and maximum diagonal line length (L(MAX)), were highly effective in detecting microcracks, as their values significantly deviated from the reference level. These results were compared with conventional statistical analysis, and interestingly, the recurrence methods demonstrated greater sensitivity, successfully detecting additional very small cutting force jumps that conventional statistical methods could not identify.