Optimization of process parameters of three-dimensional needled preforms for C/C-SiC composites

• Main Authors: QI Yun-chao, FANG Guo-dong, LIANG Jun, XIE Jun-bo
• Format: Article
• Language: zho
• Published: Journal of Materials Engineering 2020-01-01
• Series: Journal of Materials Engineering
• Subjects: needling composites; bp neural network; stiffness prediction; genetic algorithm; process optimization
• Online Access: http://jme.biam.ac.cn/CN/Y2020/V48/I1/27
• ISSN: 1001-4381; 1001-4381

A surrogate model was established to optimize needling process parameters of three dimensional needled C/C-SiC composites by using back propagation (BP) neural network and improved genetic algorithm. The relationship between needling process parameters and composites stiffness was obtained. The stiffness prediction obtained by BP neural network is in good agreement with the finite element calculated results. The maximum error of training data is 0.526%, and the maximum error of test data is 0.454%. Thus, the BP neural network model exhibits the high prediction accuracy. The genetic and optimization strategies of genetic algorithm were improved to optimize the needling process parameters. The calculated needling process parameters by the model can significantly improve the stiffness of the C/C-SiC composites. The in-plane tensile modulus increase by 11.07% and 11.48%, and the out-of-plane tensile modulus increase by 49.64% and 48.13%, respectively. The comprehensive stiffness performance of composite material increase by 18.17% and 18.21%, respectively.