Autonomous intelligent additive manufacturing of continuous fiber-reinforced composites: data-enhanced knowledgebase and multi-sensor fusion

Additive manufacturing (AM) are an emerging technique to generate complex structures of composite. However, the instability of the AM process leads to adverse manufacturing effects, including dimensional inaccuracies and poor mechanical performance in CFRP composites. Online monitoring and adaptive control based on autonomous cognition are suggested to ensure the accuracy of as-fabricated parts and enhance their quality upon manufacturing. In this study, a method of online monitoring and self-adaptive control based on multi-sensor fusion is proposed to predict and adjust the process parameters during AM of CFRP. The force sensor, visual camera, and thermal camera are employed to obtain the multiple signals upon manufacturing and then realize the autonomous perception, cognition, and decision features. Herein, the quantitative correlations between local defects and multi-sensing features are established to guide the decision-making of closed-loop adjustment. Besides, an empirical surrogate model between the misalignment of fiber bundles and input parameters is built to predict the proper parameters. Moreover, the temperature difference and contact force are selected as the controlled features, which are acquired using a thermal camera and a force sensor. The proposed system offers a novel framework for bolstering both the stability of the AM process and the quality of fabricated components.增材制造（Additive Manufacturing，AM）是一种用于制备复合材料复杂结构的新兴技术。然而，该工艺的不稳定性会引发诸多不利制造后果，例如碳纤维增强复合材料（Carbon Fiber Reinforced Plastic，CFRP）构件出现尺寸偏差与力学性能劣化等问题。现有研究表明，基于自主认知的在线监测与自适应控制策略，可保障制造成品的精度，并在制造过程中提升构件整体质量。本研究提出一种基于多传感器融合的在线监测与自适应控制方法，用于在碳纤维增强复合材料增材制造过程中预测并优化工艺参数。该方法通过力传感器、视觉相机与热成像相机采集制造过程中的多源信号，以实现自主感知、认知与决策功能。研究建立了局部缺陷与多传感特征间的定量关联关系，用于指导闭环调整的决策流程；同时构建了纤维束错位与输入参数间的经验替代模型，以实现最优工艺参数的预测。此外，本研究选取温差与接触力作为被控特征，分别通过热成像相机与力传感器获取对应数据。所提出的系统为提升增材制造工艺稳定性与制造成品质量提供了全新的技术框架。