【摘要】 本文应用人工神经网络原理,采用Levenberg-MarquardtBP算法,对于吸收光谱严重重叠的PAR-Cu、Co、Ni、Zn四组分显色体系同时进行含量测定。Cu、Co、Ni、Zn的平均回收率分别为100%、99%、101%、99%。实验表明,与普通BP网络、改进型BP网络和径向基网络相比,该算法具有训练速度快、预测结果准确度高等特点,和光度法结合有望成为多组分分析的有效方法之一。更多还原

【Abstract】 By means of the theory of artificial neural network and Levenberg-Marquardt back-propagation train algorithm, the four-component metal coordinate compounds of PAR-Cu, Co, Ni, Zn were determined simultaneously, in which the spectra overlapped severly. The mean recovery rate of Cu, Co, Ni, Zn were 100%, 99% , 101% and 99% individually. The results were better than those of radius network, modified BP network and common BP network in training speed and the accuracy. In conclusion, the new network spectrophotometry is a good choice for resolving multicomponent.更多还原