Feature-Driven Adaptive Wall-Thickness Compensation for Large Rotational Thin-Walled Components

Machining deformation causes local variations in wall thickness in large rotational thin-walled components, while limited access to the inner surface complicates measurement and compensation. This study uses raw points (RPs) on accessible reference regions left unmachined by the relevant operations, together with the reference thickness, to reconstruct the local inner profile. The component is partitioned according to process features, and the deviations in each partition are arranged as an axial sequence. The lowest polynomial order satisfying a maximum residual threshold is selected, and the resulting continuous compensation functions are mapped to densified numerical control (NC) toolpaths. Three adaptively compensated regions were compared with three manually adjusted regions on one aero-engine grid casing. The average regional wall-thickness range and sample standard deviation decreased by 67.9% and 73.4%, respectively. All 42 inspection points in the adaptive group met the 1.10–1.40 mm tolerance, compared with 32 points in the manual group. Data processing and NC-program regeneration after measurement took approximately 6 min, compared with at least 2 h for the manual workflow, a reduction of approximately 95%. These results support improved wall-thickness uniformity and program-preparation efficiency for the tested casing.