基于迭代初始化遗传算法的光谱波段选择及其在感冒液多组分测定中的应用

光谱数据用于多元校正时，组分间的交互作用会使部分波段与组分浓度呈非线性关系，在用偏最小二乘法(PLSR)建模前，宜作波长筛选。基于迭代初始化的遗传算法(IRGA)将运行多轮GA，递归地以上轮结果作为先验知识支持下轮的初始化，并对入选波长点的统计频率进行平滑处理，由此可高效地从全谱中选出校正性能良好的波段，筛选结果稳定。入选波段对全谱既作了适当简化，又充分保留了有效信息。再采用PLSR建模，模型更具稳健性。将该法用于感冒液的五组分测定，与全谱建模法相比，其预测性能和稳健性有显著提高。

Wavelength Interval Selection by Iteratively Reinitialized GA and Its Application to Spectrophotometric Determination of Components in Cough Syrup

Wavelength selection in PLS calibration can be used to reach two goals: improve the predictive ability and simplify the model.Iteratively reinitialized GA is a modified genetic algorithm,and it gives an initializing procedure of selecting the first candidates for every run of GA,which uses the results of previous runs as the guiding information.This algorithm can select wavelength regions instead of scattering points,which is very helpful in understanding the relevant parts of spectra.Furthermore,the continuous wavelength points make the PLS model more robust.Appling IRGA based wavelength selection to the UV-Vis spectrum of cough syrup,the result illustrates that PLS regression can greatly benefit from variable selection when used for multicomponent spectrophotometric determination.