52 / 2025-03-29 18:59:33

Temperature Prediction for Crystal Growth Process Based on Dynamic Mode Decomposition

Prediction

This research proposes a method for predicting the temperature of the molten silicon during crystal growth process based on dynamic mode decomposition technique. By constructing a two-dimensional spatio-temporal matrix for temperature change and performing singular value decomposition on it to extract the dominant modes, dynamic temperature prediction is achieved. The prediction accuracy is optimized by combining with an adaptive model update strategy. The study finds that dynamic mode decomposition can effectively capture the temperature change trend in short-term predictions. Increasing the length of the input matrix can improve the stability of long-term predictions. This method provides an efficient prediction method for optimizing the single crystal silicon growth process and offers new ideas for the temperature control of high-temperature industrial processes.