小波变换-人工神经网络用于反相高效液相色谱法同时测定靛蓝和靛玉红

APPLICATION OF ALGORITHMS OF WAVELET TRANSFORM (WT)-ARTIFICIAL NEURAL NETWORK(ANN) AND WT-ROBUST REGRESSION(RR) TO HPLC DETERMINATION OF INDIGO AND INDIRUBIN

提出了小波变换一人工神经网络（WT-ANN）和小波变换-稳健回归（WT-RR）两算法的数学模型，并应用于高效液相色谱法测定扳蓝根、大青叶及中成药复方板蓝根颗粒中靛蓝及靛玉红的测定，从而避免了繁琐的分离，实现了被测组分的直接、同时测定。所用色谱柱为ExtendC1s（4．6mm×250mm，5μm），所用流动相为甲醇与水（95＋5）的混合溶液。由测得的相对标准偏差（RSD）可知两算法均可用于试样中多组分的测定，但WT—ANN法（RSD〈0．6％）稍优于WT-RR法（RSD〈2.5％）。三种实样的分析应用中，回收率在98．8％～101．5％之间。此方法可作为工业生产中在线分析法。

Two algorithms of WT-ANN and WT-RR were modelled and applied to HPLC determination of indigo and indirubin in root and leaves of isatis and in ready-prepared Chinese Medicine called ＂Balangen＂ compound granules, making it possible to avoid the tedious separation and to determine these two components directly and simultaneously, The components to be determined were extracted from the sample with chloroform, and the extract was concentrated to a volume of 10 mL quantitatively and the solution was ready for HPLC determination, The Extend C18 （4.6 mm× 250 mm 5 μm） chromatographic column and a mobile phase of mixed solution of CHa OH and H2O （95＋5） were used in the determination. As shown by the RSD＇s obtained, both of the algorithms were satisfactory for the analysis of multi-components in a sample, but WT-ANN （RSD〈0. 6%） was better than WT-RR （RSD〈52. 5%） in the application. In the analysis of 3 substantial samples, recoveries in the range of 98. 8%- 101.5% were obtained. It was shown that the proposed method was feasible to be used as a method of on line analysis in the production-process.