摘要
分析影响钢的CCT(过冷奥氏体连续冷却转变)曲线的主要因素,基于BP神经网络算法及特征,建立CCT曲线的预测模型,并建立与之匹配的训练样本集。通过大量的实验,确定稳定的、具有预测功能的网络结构。预测结果能有效解决在无物理实验条件下,初步预测金属材料的组织、性能,为研制新钢材奠定基础。
The forecasting models were built based on algorithm and features of BP nerve network, the main influencing factors of steel＇s CCT （the austenitic continuous cooling transformation） curve were andyzed. And training stylebooks were established. The stable and useable network structure was ensured by lots of tests. Forecast results can preliminary forecast the microstructure and properties of the steel in the absence of physical experimental conditions. The method lays the foundation for the development of new steel.
出处
《热加工工艺》
CSCD
北大核心
2008年第22期85-87,共3页
Hot Working Technology
