热轧中厚板冲击功的神经网络预报模型

ANN Prediction Model for Impact Energy of Hot Rolled Plate

为了对济钢Q345B中厚板的常温夏比冲击吸收功进行预报并指导生产,采用神经元网络建立了冲击吸收功与化学成分、工艺参数、屈服强度及抗拉强度等因素间的关系预报模型。并分别采用实测与预测的强度值对夏比冲击吸收功进行了预报,预报结果和实测值吻合较好。在此基础上,将化学成分及生产工艺对冲击吸收功的影响进行了计算分析,得出了与基本物理冶金学规律一致的计算结果。因此,在给定化学成分、成品厚度和实测强度或强度指标可精确预报的条件下,所建立的模型能预报济钢Q345B热轧中厚板的常温冲击吸收功。

In order to predict the Charpy impact energy of Q345B hot rolled plate at Jinan steel, an ANN （Artificial Neural Network） impact energy model was developed considering the factors of composition, process parameters and strength of the plate. The predicted Charpy impact energy from measured and predicted strength, was in good a- greements with the measured values. Also, the effect of alloying elements and process parameters on impact energy was analyzed, obtaining the results consistent with the physicometallurgical rules. It is concluded that with given steel composition, plate thickness and measured or predicted strength the ANN model is able to predict the Charpy impact energy of hot rolled plate.