铜闪速熔炼过程操作参数预测模型及应用  

Prediction Model of Operating Parameters of Copper Flash Smelting

作　　者：谢锴[1] 米沙[1] 严兵[1] 李启[1] 

机构地区：[1]中南大学能源科学与工程学院,长沙410083

出　　处：《有色金属（冶炼部分）》2015年第5期5-9,46,共6页Nonferrous Metals（Extractive Metallurgy)

基　　金：国家高技术研究发展计划(863计划)重点项目(2009AA064603)

摘　　要：基于某厂实际铜闪速熔炼工艺和控制过程,对神经网络模型在铜闪速熔炼过程在线控制进行了研究。在分析影响溶剂率、熔炼氧单耗、反应塔总风量操作参数因素的基础上,提出一种基于BP神经网络的操作参数的预测方法,分别建立了输入向量只包含主要元素和考虑杂质元素的BP神经网络模型。网络的训练和测试结果表明,两种神经网络的输出值与实际值的最大相对误差均小于1.0%,输出值与实际样本值吻合得较好,模型输入参数中包括杂质元素时具有更高的计算精度。Based on practical copper flash smelting and control process, neural network model was studied in copper flash smelting process on-line control. Based on analysis of the effecting factors on solvent rate, oxygen consumption of melting, and air volume of reaction tower, a back propagation neural network prediction model was presented to predict the above parameters. BP neural network models with input vector containing only main elements and impurity elements in concern were established. The simulative results show that the maximum relative error between output value and actual value is less than 1.0%. Output value accords with practical data very well. The mode including impurity elements in input parameters has higher calculation accuracy.

关 键 词：铜闪速熔炼 在线控制 改进型BP神经网络 预测模型 

分 类 号：TF811[冶金工程—有色金属冶金] TP13[自动化与计算机技术—控制理论与控制工程]