芳香有机化合物多组分光谱分析的神经网络方法

Artificial neural netwerk for simultaneous spectrophotometric determination of multicomponent aromatic compounds

介绍了人工神经网络反传算法的理论和方法,利用这种算法及其改进算法对三种芳香有机化合物[a-萘酚（NP）、a-萘胺（NA）和 2,7二羟基萘（DH）]混合物的光谱数据进行了分辨解析,实现了多组分同时测定。这三种化合物的紫外光谱严重重叠,单纯用紫外光谱法不能同时定量测定,而用神经网络方法获得了良好结果,在1～5PPm浓度范围内,平均回收率为NP 101.35％,NA 98.11％,DH 102.5％。

The theory and method of neural network backpropagation (BP) algorithm are described. The BP algorithm and its tnodified form have been employed in the spectral data resplution for simultaneous determination of three aromaticcompoands——α-naphthal(NP),α-naphthylamine (NA) and 2,7-dithydroxynaphthalene(DH). The spectra of thesecompounds are heavily overlapped in UV region and can not be used for simultaneous quantitative analysis by conven-tional UV spectrophotometry. But the results obtained by neural network method are satisfactory. In the range of 1~ 5ppm, the average recoveries are NP 101. 35% ,NA 98.11% and DH 102. 5%.