光谱学与光谱分析
光譜學與光譜分析
광보학여광보분석
SPECTROSCOPY AND SPECTRAL ANALYSIS
2014年
10期
2723-2727
,共5页
张丙芳%苑立波%孔庆明%沈维政%张丙秀%刘成海
張丙芳%苑立波%孔慶明%瀋維政%張丙秀%劉成海
장병방%원립파%공경명%침유정%장병수%류성해
近红外光谱%煎炸老油%偏最小二乘法(PLS)%BP人工神经网络%主成分分析(PCA)
近紅外光譜%煎炸老油%偏最小二乘法(PLS)%BP人工神經網絡%主成分分析(PCA)
근홍외광보%전작로유%편최소이승법(PLS)%BP인공신경망락%주성분분석(PCA)
Near infrared spectroscopy%Frying oil%Partial least squares (PLS)%BP artificial neural network%Principal compo-nent analysis(PCA)
地沟油检测是我国食品安全最为关注的话题之一，它给人们的生活健康带来了极大的危害。国内现有的检测手段也仅停留在定性检测水平上，只能确定地沟油的有无，还难以进行定量检测。本实验利用近红外光谱技术与光纤传感技术相结合的新方法对勾兑混合油中地沟油的含量进行了定量分析。将煎炸老油与九三大豆油按照一定的体积比进行勾兑，共计50个样本，采集其近红外透射光谱，分别采用偏最小二乘法（PLS ）和BP人工神经网络建立了煎炸老油含量的定量分析模型，校正集决定系数分别为0.908和0.934，验证集决定系数分别为0.961和0.952，均方估计残差（RM S EC ）为0.184和0.136，预测均方根误差（RM-SEP）都为0.1116，符合应用要求，同时还结合主成分分析法（PCA）对煎炸老油与食用植物油进行了鉴别，识别准确率为100％。实验研究证明近红外光谱技术不仅可以准确快速的定性分析地沟油，还能定量的检测地沟油的含量，在油脂的检测方面具有很大的应用前景。
地溝油檢測是我國食品安全最為關註的話題之一，它給人們的生活健康帶來瞭極大的危害。國內現有的檢測手段也僅停留在定性檢測水平上，隻能確定地溝油的有無，還難以進行定量檢測。本實驗利用近紅外光譜技術與光纖傳感技術相結閤的新方法對勾兌混閤油中地溝油的含量進行瞭定量分析。將煎炸老油與九三大豆油按照一定的體積比進行勾兌，共計50箇樣本，採集其近紅外透射光譜，分彆採用偏最小二乘法（PLS ）和BP人工神經網絡建立瞭煎炸老油含量的定量分析模型，校正集決定繫數分彆為0.908和0.934，驗證集決定繫數分彆為0.961和0.952，均方估計殘差（RM S EC ）為0.184和0.136，預測均方根誤差（RM-SEP）都為0.1116，符閤應用要求，同時還結閤主成分分析法（PCA）對煎炸老油與食用植物油進行瞭鑒彆，識彆準確率為100％。實驗研究證明近紅外光譜技術不僅可以準確快速的定性分析地溝油，還能定量的檢測地溝油的含量，在油脂的檢測方麵具有很大的應用前景。
지구유검측시아국식품안전최위관주적화제지일，타급인문적생활건강대래료겁대적위해。국내현유적검측수단야부정류재정성검측수평상，지능학정지구유적유무，환난이진행정량검측。본실험이용근홍외광보기술여광섬전감기술상결합적신방법대구태혼합유중지구유적함량진행료정량분석。장전작로유여구삼대두유안조일정적체적비진행구태，공계50개양본，채집기근홍외투사광보，분별채용편최소이승법（PLS ）화BP인공신경망락건립료전작로유함량적정량분석모형，교정집결정계수분별위0.908화0.934，험증집결정계수분별위0.961화0.952，균방고계잔차（RM S EC ）위0.184화0.136，예측균방근오차（RM-SEP）도위0.1116，부합응용요구，동시환결합주성분분석법（PCA）대전작로유여식용식물유진행료감별，식별준학솔위100％。실험연구증명근홍외광보기술불부가이준학쾌속적정성분석지구유，환능정량적검측지구유적함량，재유지적검측방면구유흔대적응용전경。
In the present study ,a new method using near infrared spectroscopy combined with optical fiber sensing technology was applied to the analysis of hogwash oil in blended oil .The 50 samples were a blend of frying oil and “nine three”soybean oil according to a certain volume ratio .The near infrared transmission spectroscopies were collected and the quantitative analysis model of frying oil was established by partial least squares (PLS) and BP artificial neural network .The coefficients of determina-tion of calibration sets were 0.908 and 0.934 respectively .The coefficients of determination of validation sets were 0.961 and 0.952 ,the root mean square error of calibrations (RMSEC) was 0.184 and 0.136 ,and the root mean square error of predictions (RMSEP) was all 0.111 6 .They conform to the model application requirement .At the same time ,frying oil and qualified edible oil were identified with the principal component analysis (PCA) ,and the accurate rate was 100% .The experiment proved that near infrared spectral technology not only can quickly and accurately identify hogwash oil ,but also can quantitatively detect hog-wash oil .This method has a wide application prospect in the detection of oil .
