增量神经网络模型预报100t电弧炉终点碳、磷和温度的应用

Application of Increment Artificial Neural Network Model to Prediction of End-Point Carbon, Phosphorus and Temperature for an 100 t EAF Steelmaking

结合增量模型和神经网络模型的优点 ,提出增量神经网络模型 ,该模型特点为 :只注重系统输入量和输出量的变化 ,系统输入与输出增量的映射关系通过网络很快形成 ,网络结构简单。以废钢、铁水、装料制度、通电时间、吨钢氧耗和电耗相对于参考炉均值的增量为输入节点 ,对冶炼钢水终点温度和碳、磷进行预报。结果表明 ,当钢水终点温度和碳、磷含量的控制精度分别在± 10℃ ,± 0 .0 2 %和± 0 .0 0 4 %时 ,预报值命中率分别为 93% ,75 %和 86 %。

Combined the advantage of increment model and artificial neural network model, the increment artificial neural network model has been developed in this paper. The characteristics of the developed model are paying attention to the change of input and output quantity, mapping relationship of system input and output quantity forming quickly by network and structure of network being simple. The end-point temperature and carbon and phosphorus content are predicted by increment of scrap, hot metal, charging program, power on time, oxygen and electric power consumption per ton steel corresponding to reference furnace average value as input nodes. The results showed that as controlled precision of carbon and phosphorus content and temperature of molten steel was ±0.02%, ±0.004% and ±10 ℃ individually, the percentage of hits of predicted value was respectively 75%, 86% and 93%.