基于CS-SVM的山梨酸钾的荧光光谱检测法研究  

Study on Fluorescence Spectrometric Detection of Potassium Sorbate Based on CS-SVM

作　　者：王书涛[1] 朱彩云 刘洺辛 彭涛[1] 程琪 孔德明[1] 王玉田[1] WANG Shu-tao;ZHU Cai-yun;LIU Ming-xin;PENG Tao;CHENG Qi;KONG De-ming;WANG Yu-tian(Key Laboratory of Measurement Technology and Instrumentation,School of Electrical Engineering,Yanshan University,Qinhuangdao,Hebei 066004,China)

机构地区：[1]燕山大学电气工程学院测试计量技术及仪器重点实验室,河北秦皇岛066004

出　　处：《计量学报》2018年第5期747-752,共6页Acta Metrologica Sinica

基　　金：国家自然科学基金(61771419;61501394;61471312);河北省自然科学基金(F2017203220)

摘　　要：通过分析山梨酸钾的橙汁溶液的荧光光谱特征,发现山梨酸钾在激发波长为375 nm、发射波长为450~510 nm范围的光谱图有毛刺,说明橙汁被激发的荧光会干扰山梨酸钾的荧光光谱。构建布谷鸟搜索算法(CS)优化支持向量机(SVM)模型对15个样本进行训练,并预测7个样本的山梨酸钾的浓度。CS-SVM的平均回收率为99.07%,均方误差为1.21×10-5g/L,结果表明CS-SVM能够精确测定橙汁溶液中山梨酸钾的浓度,CS-SVM训练过程和对预测结果的平均回收率、误差都优于PSO-SVM和GA-SVM。By analyzing the fluorescence spectra of potassium sorbate in orange juice solution, it is found that fluorescence spectrum of potassium sorhate has glares where emission wavelength is in the range of 450 -510 nm when excitation wavelength is at 375 nm, which indicates that the fluorescence characteristics of orange juice could interfere with the fluorescence spectrum of potassium sorbate. The model of cuckoo search algorithm （ CS ） optimizing support vector machine （SVM） is built to train 15 samples and predict the concentration of 7 potassium sorbate samples. The average recovery rate of CS-SVM is 99. 07% and the mean square error is 1. 21 x 10 -5 g/L. The results show that the training process,the average recovery and error of CS-SVM are all better than PSO-SVM and GA-SVM, and that CS-SVM can accurately determine the concentration of potassium sorbate in orange juice solution.

关 键 词：计量学 山梨酸钾 荧光光谱 布谷鸟搜索算法 支持向量机 

分 类 号：TB99[一般工业技术—计量学]