有色金属科学与工程  2011, Vol. 2 Issue (1): 5-8
 文章快速检索 高级检索
 基于自适应模糊神经网络的铜闪速熔炼渣含Fe/SiO2模型研究 [PDF全文]

1. 江西理工大学材料与化学工程学院, 江西 赣州 341000;
2. 中南大学冶金科学与工程学院, 长沙 410083

Research of the Fe/SiO2 in Slag Model of Copper Flash Smelting Process Based on ANFIS
ZENG Qing-yun1,2, WANG Jin-liang1,2, ZHANG Chuan-fu2
1. Faculty of Material and Chemical Engineering, Jiangxi University of Science and Technology, Ganzhou 341000, China;
2. Central South University School of Metallurgical Science and Engineering, Changsha 410083, China
Abstract: The Fe/SiO2 in Slag model of copper flash smelting process, which has the net-structure of 3 in-put, single out-put data, and the membership functions are [5 3 5], was developed based on Adaptive Fuzzy Inference System and the practical data from one Copper smelter. The results indicate the average absolute error of train samples is 0.0055 and the relative error percentage is 1.4%. The simulation results show that the average absolute error is 0.028%, and the relative error percentage is 2.9%. It means that the simulative results accord to the practical data. Thus, the model has good reference value on process optimized control of copper smelting. It also can be used in industrial online control to replace the model of static mixture.
Key words: copper    flash smelting    fuzzy control    neural network    simulatin
0 引言

1 基于自适应模糊神经网络推理系统的建模原理

 图 1 Sugeno模糊系统等效的ANFIS结构

2 铜闪速熔炼渣含Fe/SiO2 ANFIS模型的建立 2.1 铜闪速熔炼过程

2.2 原始数据处理

2.3 铜闪速熔炼渣含Fe/SiO2 ANFIS模型的MATLAB实现

MATLAB的模糊逻辑工具箱只能支持一阶或零阶的Sugeno系统和单输出系统，并且采用权重平均法，解模糊化及所有规则取单位权重1.对于解耦的多输出系统，可看成是多个单输出系统的简单叠加.

fismat= genfis2(datin，datout，0.5);

[fismat3，trnErr，stepSize，fismat4，& kErr]=anfis

([trndatin，…，trndatout]，fismat，[epoch_n 0 0.1], [],

[chkdatin chkdatout])

outfismat= evalfis(trndatin.fismat3)

 图 2 经减类聚法训练后得到的网络结构

3 仿真结果与分析

 图 3 ANFIS网络训练结果

 图 4 ANFIS模型检验数据检验结果

4 结论

 [1] 周旦荣. 世界首创的镍闪速炉计算机在线控制[J]. 有色冶炼, 1996, 25(A01): 50–53. [2] 曾青云. 铜闪速熔炼操作数据的回归分析[J]. 有色金属:冶炼部分, 1994(5): 4–6. [3] 曾青云, 汪金良. 铜闪速熔炼神经网络模型的建立[J]. 南方冶金学院学报, 2003, 24(5): 15–18. [4] 曾青云, 周立, 汪金良, 等. 基于自适应模糊神经网络的铜闪速熔炼冰铜温度模型研究[J]. 有色金属:冶炼部分, 2007(2): 2–4. [5] 汪金良, 卢宏, 曾青云, 等. 基于遗传算法的铜闪速熔炼过程控制优化[J]. 中国有色金属学报, 2007, 17(1): 156–159. [6] WANG Jin-lung, ZHANG Chuan-fu, ZENG Qing-yun, et al. Modeling and Optimization of Copper Flash Smelting Process Based on Neural Network[J]. The Chinese Journal of Process Engineering, 2008, 8(6): 105–109. [7] 汪金良, 卢宏, 汪仁良, 等. 铜闪速熔炼影响规律的神经网络分析[J]. 有色金属:冶炼部分, 2008(2): 2–4. [8] Li Zhen -Quan, Kecman V, Ichikawa A. Fuzzified Neural Network Based on Fuzzy Number Operation[J]. Fuzzy Sets and Systems, 2002, 130: 291–304. DOI: 10.1016/S0165-0114(01)00229-9. [9] Ishibuchi H, Nii M. Numerical Analysis of the Leaning of Fuzzified Neural Networks from Fuzzy IF -THEN Rules[J]. Fuzzy Sets and Systems, 2001, 120: 281–307. DOI: 10.1016/S0165-0114(99)00070-6. [10] Akhmetov, Dote DF, Ovaska Y, et al. Fuzzy Neural Network with General Parameter Adaptation for Modeling of Nonlinear Time -series[J]. IEEE Transactions on Neural Networks / A Publication of the IEEE Neural Networks Council, 2001, 12(1): 148–152. DOI: 10.1109/72.896803.