人工神经网络解析分光光度法同时测定钨和钼

用单纯形法优化了钨、钼－二溴羟基苯基荧光酮－ＣＴＭＡＢ显色体系的实验条件。应用三层ＡＮＮ－ＢＰ网络解析钨和钼的吸收光谱，分光光度法同时测定了钨和钼并与偏最小二乘法、因子分析、Ｐ－矩阵法、卡尔曼滤波、主成分回归等化学计量学方法的解析结果进行了比较，表明神经网络方法优于其它方法。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。

Simultaneous Determination of Tungsten and Molybdenum by Artificial Neural Networks Spectrophotometry

In this paper a three-layer artificial neural network (ANN) has been applied to the simultaneous determination of tungsten and molybdenum by spectrophotometry. The results obtained with ANN are better than that with other methods of chemometrics as partial least squares, factor analysis, P-matrix method, Kalman filtering methods, and principal component regression. The conditions of simultaneous determination by color reaction of W (VI), Mo (VI) with dibromo-hydroxyphenyl fluorone and cetyltrimethylammounium bromide have been optimized by simplex method. A method has been developed for simultaneous determination of tungsten and molybdenum in steel with satisfactory results. The paralysis in the procedure of training about ANN has been avoided with the improved backpropagation (BP)algorithms. The simple transformation method for target vectors and criterion function convenient for selection of the parameters about the network has been put forward.