论文部分内容阅读
高炉炼铁通常采用铁水Si含量间接反映炉温的变化,模型预测精度低。以影响炉温的6个变量为输入变量,采用基于自组织的分布式RBF神经网络模型分别对铁水温度和铁水Si含量建立预测模型,先用自组织神经网络划分输入输出样本空间,然后对每个子空间建立RBF神经网络子网模型,再使用子网模型对测试样本集的同一个样本点进行预测,并以测试样本点对每一子空间的隶属度为权值,对子网预测值进行加权求和,得到最终预测值。对比使用同一输入变量数据的铁水温度和铁水Si含量的预测模型命中率,研究表明,高炉铁水温度的命中率更高,具有更好的炉温预测效果。
Blast furnace iron smelting iron content is usually used to indirectly reflect the change of furnace temperature Si, model prediction accuracy is low. Taking the 6 variables that affect the furnace temperature as input variables, a self-organizing distributed RBF neural network model was used to establish a prediction model for the hot metal temperature and hot metal Si content respectively. The input and output sample space was divided by self-organizing neural network, Sub-space RBF neural network sub-network model is established, and then the sub-network model is used to predict the same sample point of the test sample set. The sub-network sub-network membership is taken as the weight of the test sample points. Weighted sum, get the final predictive value. Comparing the predictive model hit rates of hot metal and hot metal Si using the same input data, the research shows that hot metal hot blast has a higher hit rate and better predicting effect of temperature.