Empirical Investigation and Feed-Forward Control of Wire Tension in Needle Winding Processes

In the evolving landscape of electric motor manufacture, needle winding technology has emerged as a viable solution, offering precision in producing orthocyclic round wire windings. The practicality of this technology, however, depends significantly on stable wire tension during the winding process. To facilitate this stability, a comprehensive model describing the forces exerted on the wire at various winding stages becomes indispensable. This paper primarily focuses on developing a model-based feed-forward wire brake torque control profile. This new approach ensures steady wire tensile forces at the wire deposit point, presenting practitioners with an alternative to or enhancement of feedback control systems, avoiding the need for intricate wire tension measurements in production. Leveraging previous research, we utilize neural networks to approximate the bending forces at the winding needle while using a simplified linear model for the preceding wire guiding system. Further, through meticulous analysis of measured tensile force signals, we introduce an empirical model to approximate the wire’s dynamic force variations. Force measurements were then used to evaluate the generated model-based braking torque profile. The findings revealed a reduction of measured median absolute deviation by more than $\mathbf{5 0 \%}$, demonstrating effective stabilization of the wire tension compared to a constant brake torque control.