The ceramic fiber textile molding process plays a crucial role in the production of high-temperature insulation materials, where molding accuracy and defect control directly impact product performance. Traditional processes rely on experience and manual adjustments, making it difficult to achieve efficient and precise optimization and defect suppression. Existing optimization methods often focus on a single scale or local defect control, lacking comprehensive analysis and multiscale optimization approaches, and are also limited by the absence of effective AI models for defect prediction and suppression. This paper proposes an optimization method for the ceramic fiber textile molding process based on a multiscale AI model. By integrating deep learning and data-driven multiscale modeling techniques, a model is developed that can simulta-neously process macroscopic textile parameters and microscopic defect features, enabling real-time optimization and defect prediction during the textile process. Experimental results show that the proposed method achieves quantifiable improvements in molding accuracy and defect rate, with accuracy increased by approximately 30%-35% and defect rate reduced by about 32%-38%. This research provides a new approach to the optimization of ceramic fiber textile molding processes, effectively improving production efficiency and product quality. It holds considerable academic and practical value and shows promising application prospects in the production of high-temperature insulation materials.

ceramic fiber, textile molding process, multiscale AI model, process optimization, defect suppression