基于神经网络的铜闪速熔炼过程工艺参数预测模型-中南大学学报.PDFVIP

第 38 卷第 3 期 中南大学学报( 自然科学版) Vol.38 No.3 2007 年 6 月 J. Cent. South Univ. (Science and Technology) Jun. 2007 基于神经网络的铜闪速熔炼过程工艺参数预测模型 喻寿益，王吉林，彭晓波 ( 中南大学 信息科学与工程学院，湖南 长沙，410083) 摘 要：基于铜闪速熔炼过程是典型的高温、多相多组分复杂生产过程，熔炼时，闪速炉内发生激烈而迅速的化 学反应，冰铜品位、冰铜温度和渣中铁硅比是铜熔炼过程的关键三大工艺参数，在线检测时存在成本高、滞后大， 实现困难等问题，在分析影响工艺参数因素的基础上，提出一种基于 BP 神经网络的三大工艺参数预测方法，通 过收集现场生产数据，挖掘其中隐含的工艺参数信息，建立预测模型。仿真结果表明，这三大工艺参数的最大绝 对误差分别为 0.630，6.680 和 0.051，最大相对误差分别为 1.16%，0.55%和 3.40%，说明模型预测结果与实际生 产数据较吻合，该预测模型可用来指导实际生产操作，并可用于铜闪速熔炼过程参数优化。 关键词：铜闪速熔炼；BP 神经网络；预测模型 中图分类号：TP13 文献标识码：A 文章编号：1672-7207(2007)03−0523−05 Prediction model of craft parameters based on neural network during the process of copper flash smelting YU Shou-yi, WANG Ji-lin, PENG Xiao-bo (School of Information Science and Engineering, Central South University, Changsha 410083, China) Abstract: Copper flash smelting is a complex industrial process with multiple phases and multiple components at high temperature. During the process of smelting, the reactions occur fiercely and rapidly in flash furnace, and matte grade, matte temperature and mass rate of Fe and SiO2 are the three key technology parameters during the process of copper flash smelting. The measurement method of these parameters is not only hard to be detected on-line, but also has time-delaying and costs a lot. A back propagation neural network prediction model was presented to predict these parameters, and its simulation experiment was given. The simulative results show that the biggest absolute error of the three parameters are 0.630,