光电工程
光電工程
광전공정
OPTO-ELECTRONIC ENGINEERING
2015年
2期
47-51
,共5页
毕卫红%高明明%付兴虎%吴国庆%陈俊刚%于腾飞
畢衛紅%高明明%付興虎%吳國慶%陳俊剛%于騰飛
필위홍%고명명%부흥호%오국경%진준강%우등비
藻红蛋白%可见光吸收光谱%偏最小二乘法%主成分回归法
藻紅蛋白%可見光吸收光譜%偏最小二乘法%主成分迴歸法
조홍단백%가견광흡수광보%편최소이승법%주성분회귀법
phycoerythrin%visible absorption spectrometry%partial least squares%principal component regression
藻红蛋白(Phycoerythrin, PE)是衡量水体中藻类富营养化的一个重要指标,可见光吸收法具有测量快速、准确、无二次污染、可实现在线和原位测量等优点。因此,本文提出采用可见光吸收法测定水中藻红蛋白含量。以藻红蛋白标准液为研究对象,采集5~250 mg/L不同浓度标准液的可见光吸收光谱,对藻红蛋白进行定性定量分析。同时,对光谱的不同谱区预处理后,运用偏最小二乘法(PLS)建立可见光吸收光谱下的藻红蛋白校正分析模型,并与主成分回归法(PCR)进行比较。实验结果表明,采用PLS算法在谱区480~570 nm建立的校正分析模型的相关系数R2可达0.993,校正标准差RMSEC与相对标准差RSD最小,分别为5.5783和3.92%。
藻紅蛋白(Phycoerythrin, PE)是衡量水體中藻類富營養化的一箇重要指標,可見光吸收法具有測量快速、準確、無二次汙染、可實現在線和原位測量等優點。因此,本文提齣採用可見光吸收法測定水中藻紅蛋白含量。以藻紅蛋白標準液為研究對象,採集5~250 mg/L不同濃度標準液的可見光吸收光譜,對藻紅蛋白進行定性定量分析。同時,對光譜的不同譜區預處理後,運用偏最小二乘法(PLS)建立可見光吸收光譜下的藻紅蛋白校正分析模型,併與主成分迴歸法(PCR)進行比較。實驗結果錶明,採用PLS算法在譜區480~570 nm建立的校正分析模型的相關繫數R2可達0.993,校正標準差RMSEC與相對標準差RSD最小,分彆為5.5783和3.92%。
조홍단백(Phycoerythrin, PE)시형량수체중조류부영양화적일개중요지표,가견광흡수법구유측량쾌속、준학、무이차오염、가실현재선화원위측량등우점。인차,본문제출채용가견광흡수법측정수중조홍단백함량。이조홍단백표준액위연구대상,채집5~250 mg/L불동농도표준액적가견광흡수광보,대조홍단백진행정성정량분석。동시,대광보적불동보구예처리후,운용편최소이승법(PLS)건립가견광흡수광보하적조홍단백교정분석모형,병여주성분회귀법(PCR)진행비교。실험결과표명,채용PLS산법재보구480~570 nm건립적교정분석모형적상관계수R2가체0.993,교정표준차RMSEC여상대표준차RSD최소,분별위5.5783화3.92%。
Phycoerythrin (PE) is an important indicator which measures the algae eutrophication in water. Visible absorption spectrometry has some advantages, such as rapid detection, accurate, no secondary pollution, on-line and in-situ measurements etc. Therefore, the method of detection phycoerythrin contents by visible absorption spectrometry is proposed. Based on the standard solution of phycoerythrin as experimental subjects, the visible absorption spectrometry of the standard solution are collected in the range of 5 ~ 250 mg/L, and the correction analysis model is made. The spectral preprocessing is performed in different spectral region and the Partial Least Squares (PLS) algorithm is used to establish the correction analysis model. Compared with Principal Component Regression (PCR) algorithm, the results show that the correction analysis model has the highest correlation and smallest error by the PLS algorithm made in the spectral region of 480~570 nm. The correlation coefficient root mean square error of calibration and relative standard deviation is 0.9930, 5.5783 and 3.92%, respectively.