An adaptive cooperative swarm optimization method for noise-robust cylindricity evaluation in signal processing

Accurate cylindricity evaluation is imperative for ensuring dimensional integrity and process stability in precision manufacturing. However, high-density measurement signals are often affected by outliers, noise, and multi-source uncertainty, leading to biased assessments. The present study proposes an Adaptive Cooperative Particle Swarm Optimization (ACPSO) framework for robust Minimum-Zone Cylindricity (MZCY) evaluation. The method integrates local-region outlier suppression, adaptive parameter adjustment, and axial-direction constraint into a unified optimization scheme. The measurement data are segmented into angular subregions, and percentile-based thresholds are applied to remove abnormal signals. The refined data are then processed through a cooperative multi-swarm optimization, where adaptive learning parameters dynamically balance exploration and convergence. The findings, which are supported by simulation and experimental results, including comparisons with APSO, PSO, and RPM-APSO, and validation using a Taylor Hobson Talyrond-2000, demonstrate that ACPSO achieves higher precision, lower uncertainty, and better robustness. The proposed framework provides an efficient and reliable approach for high-fidelity cylindricity assessment in intelligent metrology.