A framework for cross-evaluation driven by data integration of TANH-FEM and YOLOv11-OBB for the optimization of Ti6Al4V machining

High-speed machining of Ti6Al4V commonly produces serrated chips and periodic cutting-force fluctuations, which may increase cyclic loading and tool-fatigue risk. Conventional local node-tracking and manual geometric measurement can become labor-intensive and difficult to scale for severely distorted chip meshes, limiting rapid process evaluation. This study develops a cross-evaluation framework integrating TANH-based finite element modeling with YOLOv11-OBB machine vision to quantify simulated chip serration automatically and examine its relationship with cutting-force fluctuations. A two-dimensional plane-strain orthogonal-cutting model was established, and the TANH constitutive relation was implemented through a VUMAT subroutine to account for strain-softening behavior under high-strain-rate cutting conditions. Four cutting speeds of 80, 100, 120, and 140 m/min were investigated. YOLOv11-OBB was then used to identify chip peak and valley regions in finite element contour plots. The cutting conditions were cross-evaluated using serration degree and cutting-force fluctuation amplitude. The YOLOv11-OBB detector achieved an [email protected] of 0.779 and a maximum F1 score of 0.73 at a confidence threshold of 0.19. The mean serration degrees at the four speeds were 0.450, 0.421, 0.419, and 0.437, respectively, while the corresponding force-fluctuation amplitudes were approximately 650, 681, 582, and 743 N. Both quantified responses varied non-monotonically as cutting speed increased. Among the four conditions, 120 m/min produced both the lowest serration degree and the lowest force-fluctuation amplitude, indicating a comparatively favorable response within the investigated range. The proposed framework enables efficient batch post-processing of finite element results and provides a numerical approach for preliminary screening of Ti6Al4V machining conditions within the present simulation framework.