Polymer glass transition temperature prediction method and device based on adaptive optimization algorithm and graph neural network, and medium
By combining an improved stochastic gradient optimization algorithm with a graph neural network, dynamically adjusting the learning rate and conducting phased training, the problems of low efficiency and insufficient accuracy in the existing technology for predicting the glass transition temperature of polymers are solved, and efficient and accurate polymer Tg prediction is achieved.
Patent Information
- Application Number
- CN202510689376.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-27
- Publication Date
- 2025-09-12
AI Technical Summary
Existing technologies are inefficient, inaccurate, limited in applicability, and have poor model interpretability when predicting the glass transition temperature of polymers, making it difficult to meet the needs of high-throughput material research and development.
Combining the improved stochastic gradient optimization algorithm (RAdam) with multi-scale molecular feature modeling, by constructing a graph neural network model, dynamically adjusting the learning rate, performing phased training, and using a gradient accumulation strategy and validation set evaluation, the model performance is optimized.
It achieves efficient and accurate prediction of the glass transition temperature of polymers, improves the training efficiency and generalization ability of the model, adapts to different polymer systems, reduces computational complexity, and ensures the stability and reliability of training.