BP人工神经网络分光光度法同时测定铜精矿中铅和锌  

Simultaneous determination of lead and zinc in copper concentrate by BP-artificial neural network spectrophotometry

作　　者：于建忠[1] 于凯妍[2] 史晓燕[1] 彭芮[1] 

机构地区：[1]天津职业大学,天津300402 [2]南开大学信息技术科学学院,天津300071

出　　处：《冶金分析》2009年第3期52-55,共4页Metallurgical Analysis

摘　　要：在不分离铜基质等主成分下，探讨了pH4．4、TritonX-100存在的条件下，以2-[（5-溴-2吡啶）-偶氮]-5-二乙氨基酚（5-Br—PADAP）为显色剂，BP人工神经网络分光光度法同时测定铜精矿中铅和锌的分析方法。实验采用三层前馈反向传播人工神经网络。隐层选择节点数为8，采用S型神经元；输出层选择节点数为2，采用线性神经元。网络经过555次训练即可达到指定的误差域值。铅和锌的质量浓度分别在0～1．2μg／mL和0～1．4μg／mL服从比尔定律。该方法已用于铜精矿中铅、锌的测定，相对标准偏差分别为0．8％和1．4％；回收率为89．3％～109．6％和99．2％～107．3％。Simultaneous determination of lead and zinc in copper concentrate by BP artificial neural network spectrophotometry was investigated without separation of matrix copper using 2-[（5-bromine-2- pyridine）-azo]-5- diethylaminophenol （5-Br-PADAP） as color reagent in the presence of TritonX-100 at pH 4. 4. Triple feedforward antipropagation artificial neural network was used in this study. For the hidden layer, eight nodes were selected and the S-type nerve unit was adopted, whereas for the output layer,two nodes were selected and the linear nerve unit was adopted. The specified error range could be reached after 555 times of learning. Beer＇s law was obeyed in the mass concentration of 0-- 1.2 μg/mL and 0--1.4 μg/mL for lead and zinc. This method has been applied to the determination of lead and zinc in copper concentrate with the relative standard deviations of 0.8 % and 1.4 %, respectively and the recoveries for lead and zinc of 89.3%-109.6% and 99.2%-107. 3%.

关 键 词：BP人工神经网络 分光光度法 同时测定 铅 锌 铜精矿 

分 类 号：P575.4[天文地球—矿物学]