Digital Twin-Based Process Monitoring, In-Situ Defect Detection, and Real-Time Process Adjustment Solution for Automated Fiber Placement (AFP) Process

Automated fiber placement (AFP) machines are currently used by aerospace manufacturers for the manufacturing of fuselage components and other composite aerospace structures. Despite their considerable use in the aerospace industry, the AFP process requires time-consuming manual inspections by the operator, and process induced defects can result in significant re-work. As an example, in the AFP process, approximately 42% of the work time is spent on inspection and data review, while only 19% of the time is spent on the actual lay-up of composite material (productive portion). This results in the AFP process having longer production times, more rework and lower overall productivity than desired. Predictronics, in collaboration with the University of Delaware Center for Composite Materials (CCM) is proposing a digital twin-based process monitoring, in-situ defect detection, and real-time process adjustment solution to address the previously mentioned challenges regarding lower productivity due to significant manual inspection, high amounts of rework, and quality-related scrap. The digital would be demonstrated using a state-of-the-art Mikrosam AFP machine at the CCM facilities. Existing sensors and data would be considered from the AFP machine, including an IR camera, pyrometer (temperature measurement), and existing tags from the control system (such as laser power set-point and actual values, etc.). The digital twin would utilize a combined image and sensor-based in-situ defect detection model; in which phase I would demonstrate the feasibility of this model with offline results and a dashboard demonstration (using Predictronics PDX software and dashboard). The digital twin would also combine multiple data-driven machine learning models that use various machine settings and inputs to predict the composite part temperature. The feasibility in phase I for this digital twin technology would provide the foundation for expanding this capability to cover more defect types and allow for real-time process adjustments for the phase II effort.