Інтелектуальне керування процесами намотування нитки на верстатах з ЧПК на основі цифрових двійників

The increasing complexity of composite manufacturing processes requires advanced control strategies that ensure high precision and adaptability of CNC-based filament winding systems. This paper proposes an intelligent approach to the control and optimization of filament winding operations based on the integration of digital twin technology and machine learning methods. The study focuses on the development of a digital twin of a two-axis filament winding machine that replicates both kinematic behavior and process-related phenomena, including fiber placement accuracy and servo drive dynamics. Unlike conventional approaches, the proposed framework incorporates predictive models of servo error that account for higher-order motion characteristics such as acceleration and jerk. These models are embedded into the digital twin environment, enabling real-time estimation and compensation of trajectory deviations. A machine learning module is introduced to identify nonlinear dependencies between control inputs and resulting winding errors based on experimental data. The trained model is used to adjust CNC command signals adaptively, ensuring improved trajectory tracking accuracy under varying operating conditions. Experimental validation was carried out on a laboratory-scale winding machine using fiberglass rovings. The results demonstrate that the proposed intelligent control approach reduces positioning error by up to 65% compared to conventional control strategies. Additionally, the digital twin enables rapid parameter tuning and reduces setup time for new winding configurations. The developed methodology provides a foundation for the creation of smart manufacturing systems in composite production, combining digital twins, predictive modeling, and adaptive control.