基于粒子群优化的波段选择方法在多组分同时测定中的应用  

Simultaneous Determination of Multicomponent by Wavelength Selection Using Particle Swarm Optimization Algorithm

作　　者：陶丘博[1] 申琦[1] 张小亚[1] 刘红[1] 

机构地区：[1]郑州大学化学系,郑州450001

出　　处：《分析化学》2009年第8期1197-1200,共4页Chinese Journal of Analytical Chemistry

基　　金：国家自然科学基金(No.20505015)资助项目

摘　　要：多组分同时测定时,组分间发生相互作用,应作波段选择。提出了改进的粒子群优化算法进行随机的波段选择。所提出的方法用于邻、间、对硝基苯酚的浓度预测,在208～481nm范围内,以0.15mol/LNaOH为溶剂,配置27组混合液作训练集,27组作预测集。训练集的均方根误差(RMSE)分别为0.1257、0.2228和0.0846;预测集RMSE分别为0.2070、0.1507和0.394,得到了较好的预测结果。The problem of interaction of components and determination of one analyte in the presence of other compounds could be solved by selecting wavelength. The wavelength interval selection by modified particle swarm optimization algorithm was suggested for spectrophotometric determination. The proposed approach was used to determine the content of nitrophenol isomers. In the range of 208 -481 nm, 27 different mixtures of nitrophenol isomers in NaOH solution （0.15 moL/L） were used as calibration set, and the remaining 27 mixtures were used as test set. The root mean square errors （RMSE） were 0. 1257, 0. 2228 and 0. 0846 for training set respectively. The root mean square errors of prediction were 0. 2070, 0. 1507 and 0. 3940 for o-nitrophenol, m-nitrophenol and p-nitrophenol, respectively. The proposed method shows good prediction performance.

关 键 词：粒子群优化算法 波段选择 硝基苯酚 

分 类 号：O651[理学—分析化学] TP301.6[理学—化学]