用反向传播人工神经网络预测低碳低合金钢的马氏体转变开始温度

PREDICTING THE MARTENSITIC TRANSFORMATION START TEMPERATURE USING BACK-PROPAGATION ARTIFICIAL NEURAL NETWORKS

根据收集和整理的实验数据,建立了低碳低合金钢的成分与马氏体转变开始温度(M_s点)的反向传播(BP)人工神经网络,用这种方法预测了一些钢的M_s点,并与用其它经验公式得到的结果进行了比较,结果表明:用人工神经网络能更精确地预测钢的M_s点,预测精度明显高于其它线性经验公式,另外用正交实验法设计了几种基准成分的钢,用人工神经网络分析了几种合金元素对M_s点的定量影响,计算结果表明,与传统的经验公式表达的信息不同,合金元素的含量与钢的M_s点间表现为非线性关系,可以认为,这种非线性关系是由合金元素间复杂的交互作用引起的。

The back-propagation artificial neural network was established using data collected from domestic and foreign literatures and the M-s temperatures of some steels were predicted by using the network and compared with those acquired from other methods. Results indicate that the M-s temperatures can be predicted more accurately using artificial neural networks. Moreover, the influence of alloying elements on M-s temperatures was analysed quantitatively using artificial neural networks. The results show that there exists nonlinear relationship between contents of alloying elements in steels and their M-s temperature which is related to the interaction among the alloying elements.