论文部分内容阅读
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。
The simplex method was used to optimize the experimental conditions of tungsten, molybdenum-dibromohydroxyphenylfluorone-CTMAB system. The absorption spectra of tungsten and molybdenum were analyzed by the three-layer ANN-BP network. The spectrophotometric method was used to simultaneously determine tungsten and molybdenum and the chemometrics of partial least-squares, factor analysis, P-matrix, Kalman filtering and principal component regression The analytical results of the method are compared, indicating that the neural network method is superior to other methods. Using the improved BP algorithm, the possible paralysis phenomenon in neural network learning can be avoided. A simple transformation method of target vector and convergence evaluation function which is convenient for network parameter selection are proposed.