Deep learning-based correlation analysis of ceramic fiber composites enabled by an automated manufacturing process

This study develops a material design framework for ceramic fiber reinforced composites by implementing an automated manufacturing process capable of generating the large-scale experimental dataset required for AI-based predictive models. A fully integrated continuous production line was constructed, in which key processing parameters were digitally controlled and recorded. Using this system, more than 600 experimental data points were collected and used in a deep-learning-based correlation analysis to quantify the relationships between process variables and mechanical properties. Compared with manual fabrication, the automated process significantly reduced resin content fluctuations and internal porosity, thereby enhancing structural integrity; microstructural observations further confirmed that improved fiber distribution uniformity and the suppression of resin-rich regions contributed to superior mechanical performance. Overall, the proposed framework — combining automated data acquisition with AI-based predictive modeling — provides practical design guidelines for industrial composite manufacturing and offers a foundation for future data-driven materials platforms and process optimization strategies.