Zufeng Shang, Shuo Li, Qinchuan Li, Bo Yan, Haotian Ling
Abstract The demand for lightweight and reliable composite structures in aerospace and transportation has increasingly driven the adoption of circular braiding technology. The circular braiding technique has been recognized as a key preforming process for manufacturing large components with complex geometries because of its exceptional contour adaptability, high efficiency, and superior mechanical performance. However, manufacturing components with extreme geometries, such as non-axisymmetric variable cross section and spatially curved centerlines coupled with complex topology, presents significant challenges related to physical interactions and scale effects, severely constraining process accuracy and reliability. This paper systematically proposes a four-level classification system (Class I–IV) based on geometric topology, establishing a foundational framework for assessing the adaptability of braiding processes. The applicability and limitations of analytical, kinematic, and finite element methods (FEMs) in braiding process modeling are quantitatively compared. To address the difficulty in Class IV geometries, the potential of emerging solvers like differentiable physics engines and the material point method (MPM) is explored. At the manufacturing level, challenges induced by scale effects, such as system dynamic coupling, nonlinear friction, and tension fluctuations, require innovative solutions, including novel hybrid robots, active tension control, and magnetic levitation drives. Furthermore, this work outlines a strategic route based on intelligent manufacturing involving the deep integration of digital twins and artificial intelligence to facilitate a paradigm shift from “geometry-driven” to “performance-driven” manufacturing.