论文部分内容阅读
简述炼钢冶炼工艺的特点及其复杂性,分析了冶炼工艺的决策过程,建立了转炉冶炼工艺中硅铁、硅锰合金加入量的神经网络(NN)模型。针对冶炼工艺数据的特点,研究了神经网络输入输出数据的预处理方法,引入基于对数变换的数据预处理算法,测试表明效果较优。所建立的NN模型能真实反映转炉冶炼的工艺特点,模型值和实测值相差较小,能实现在给定成品成分要求下合金加入量的自动选择。
The characteristics and complexity of the smelting process are briefly introduced. The decision-making process of the smelting process is analyzed, and a neural network (NN) model for the addition of ferrosilicon, silicon-manganese alloy in the converter smelting process is established. Aiming at the characteristics of smelting process data, the preprocessing method of input and output data of neural network is studied, and the data preprocessing algorithm based on logarithm transform is introduced. The test shows that the method has better performance. The established NN model can truly reflect the characteristics of the converter smelting process, the model value and the measured value of the difference is small, can be achieved under the requirements of a given composition of the alloy dosage automatic selection.