| description abstract | In the realm of large format additive manufacturing (LFAM), determining an effective printing strategy before actual printing involves predicting temperature behaviors and controlling layer time, which has consistently been challenging. Currently, temperature prediction for controlling layer time in LFAM is primarily conducted through offline simulations or online monitoring. However, these approaches are typically tailored to specific cases and lack generalizability. Consequently, there exists a significant gap in the development of a universal model that can leverage historical data to predict temperature across various new geometries and positions. In this article, a novel approach to optimize printing strategies for LFAM is proposed through the development and application of a transformer-based model focused on the dynamic prediction and management of temperature profiles across the print surface. Subsequently, the authors input the predicted temperature into an optimization model to determine the optimal layer time. A series of experiments were conducted to validate the effectiveness of the proposed model. Using historical temperature data collected from the real printing processes, the model demonstrated a high degree of accuracy in predicting temperature profiles for new design, enabling the optimization of layer time settings far beyond the capabilities of traditional fixed-time methods. This process significantly enhances the printing strategy, thereby increasing both the efficiency of the printing process and the quality of the printed objects. | |
