人工神经网络吸光光度法同时测定钛锆

SIMULTANEOUS DETERMINATION OF TITANIUM AND ZIRCONIUM BY ARTIFICIAL NEURAL NETWORKS ABSORPTIOPHOTOMETRY

在钛,锆-二溴羟基苯基荧光酮-CTMAB显色体系中,应用三层ANN—BP网络解析钛和锆的吸收光谱,不经分离吸光光度法同时测定钛和锆。钛和锆的表观摩尔吸光系数分别为ε_(547)=9.66×10~4,ε_(533)=1.80×10~5。同时测定合金钢中钛和锆,结果满意。使用改进的BP算法,避免了神经网络学习过程中可能产生的麻痹现象。提出了目标向量的简单变换方法及便于网络参数选择的收敛评价函数。

It is reported in this paper, a three-layer artificial neural networks was applied to the simultaneous determination of titanium and zirconium by absorptiophotometry. The optimum conditions of the color reactions of Ti( Ⅵ ), Zr( Ⅵ ) with dibromo-hydrophenylfluorone in the presence of CTMAB was studied. The apparent molar absorptivity of the complex of titanium and zirconium are 9.66×10~4L·mol^(-1)·cm^(-1) and 1.80×10~5L·mol^(-1) ·cm^(-1)respectively, A method was developed for simultaneous determination of titanium and zirconium in steel with satisfactory results. The paralysis in the procedure of training about ANN was avoided with the improved backpropagation algorithms. The simple transformation method for target vectors and criterion function convenient for selection of the parameters about the netwoks was put forward.