QUALITY ESTIMATION DURING CFRTP PRESS MOLDING BY MACHINE LEARNING OF CONDITION MONITORING
Abstract
A feature of thermoplastic carbon fiber composite material (CFRTP) is that it has high productivity that enables secondary processing such as press molding. However, there is a problem that the mechanical property of the products is uneven after press molding. The cause is that the resin melted in the press mold flows under the compression load during press molding, and the fiber orientation is different from the original material. The conventional in-process quality checking method only determines whether a measured value of a pressure sensor or the like mounted on a press die exceeds a control value, and cannot predict a material flow during molding. The purpose of this research is to develop a technology that recognizes the time series change of sensor data during press forming as a pattern and determines the quality accompanying the material flow during press forming by machine learning. First, we will develop sensor data patterning technology for viscoelastic changes at each temperature and strain rate during CFRP press molding. Second, we observe the fiber orientation change inside the molded product, and model the material flow state during press molding and evaluate the mechanical properties. Third, machine learning is performed using sensor data patterns and machine characteristic data. As a result, a technology for estimating product quality from sensor data during press molding will be developed.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 11, 2021
- Source ID
- FA23862014042
Entities
People
- Yasushi Miyano
Organizations
- Air Force Office of Scientific Research
- Kanazawa Institute of Technology
- United States Air Force