【摘要】：為快速準確地測定茶油中脂肪酸含量,建立了應(yīng)用近紅外光譜技術(shù)檢測茶油中脂肪酸含量的方法。選取市售的156份茶油樣品,利用氣相色譜儀測定其脂肪酸組成及含量,同時采用近紅外光譜儀采集油樣的光譜數(shù)據(jù),并分析原始(R)光譜、SG平滑(SG)光譜和二階導(dǎo)數(shù)變換(SD)光譜與茶油中脂肪酸含量的相關(guān)性,采用偏最小二乘回歸法(PLSR)比較全光譜波段與顯著性波段對建模精度的影響,優(yōu)選出茶油中脂肪酸含量的定量檢測模型。結(jié)果表明:茶油中棕櫚酸、油酸和亞油酸含量較高,分別為4.428%~10.931%、78.036%~84.621%、7.013%~9.863%;采集的茶油近紅外光譜曲線特征變化較為明顯,光譜特征峰的位置分布于8 600~8 200、7 300~6 900、6 000~5 500、4 800~4 500和4 500~4 000 cm 1;茶油中棕櫚酸含量與R、SG光譜吸光度呈正相關(guān),油酸和亞油酸含量與R、SG光譜吸光度呈負相關(guān),SD光譜數(shù)據(jù)與棕櫚酸、油酸和亞油酸含量之間的相關(guān)系數(shù)與R和SG光譜吸光度比較,相關(guān)性極大被削弱;基于全波段建立的PLSR模型對棕櫚酸、油酸和亞油酸含量的整體預(yù)測精度略高于顯著性波段所建立的模型,校正集相關(guān)系數(shù)RC和預(yù)測集相關(guān)系數(shù)RP分別為0.837~0.956和0.818~0.938。從模型的復(fù)雜程度分析,采用顯著性波段建模的輸入變量的數(shù)量可壓縮至全波段建模的25%以下;SG PLSR模型對棕櫚酸、油酸和亞油酸含量的綜合預(yù)測性能最優(yōu),相應(yīng)的RP和預(yù)測集均方根誤差(RMSEP)分別為0.938、0.930、0.925和0.560、0.438、0.287。
[Abstract]:In order to determine the fatty acid content of tea oil quickly and accurately, a near infrared spectrometric method was established for the determination of fatty acid content in tea oil. The fatty acid composition and content of 156 tea oil samples sold on the market were determined by gas chromatograph, and the spectral data of oil samples were collected by near infrared spectrometer. The correlation between the smooth (SG) spectrum and the second derivative transform (SD) spectrum of the original (R) spectrum and the fatty acid content in tea oil was analyzed. The effects of the full spectral band and the significant band on the modeling accuracy were compared by partial least square regression (PLSR). A quantitative determination model of fatty acid in tea oil was selected. The results showed that the contents of palmitic acid, oleic acid and linoleic acid in tea oil were 4.428, 10.931and 78.03636, respectively, and the content of palmitic acid, oleic acid and linoleic acid in tea oil was 84.62113 and 7.0133.The characteristic changes of the near infrared spectrum curve of tea oil were obvious, the results showed that the content of palmitic acid, oleic acid and linoleic acid in tea oil was 4.428. The position of the spectral characteristic peak was distributed in 8 600 ~ 8 200 ~ 7 300 ~ 6 900 ~ 6 000 ~ 6 000 ~ 5 500 ~ 5 500 ~ 4 800 ~ 4 000 cm ~ (-1) and 4 500 ~ 4 000 cm ~ (-1), the content of palmitic acid in tea oil was positively correlated with the spectral absorbance of RGG, the content of oleic acid and linoleic acid was negatively correlated with the absorption of R ~ (2 +) SG and the SD spectral data were negatively correlated with palmitic acid. The correlation coefficient between the content of oleic acid and linoleic acid was greatly weakened compared with the absorbance of R and SG spectra. The overall prediction accuracy of oleic acid and linoleic acid content is slightly higher than that of the model established in the significant band. The correlation coefficient RC and the correlation coefficient RP of the corrected set and the predicted set are 0.837 ~ 0.956 and 0.818 ~ 0.938, respectively. From the analysis of the complexity of the model, the synthetic prediction performance of palmitic acid, oleic acid and linoleic acid content can be optimized by using the input variables of significant band modeling to 25% or less than 25% of the full-band modeling model, and the results show that the model can be used to predict the content of palmitic acid, oleic acid and linoleic acid. The root mean square error (RMSEP) of the corresponding RP and prediction set were 0.938 0. 930 0. 925 and 0. 560 0. 438 / 0.287, respectively.
【作者單位】： 中南林業(yè)科技大學(xué)機電工程學(xué)院;中南林業(yè)科技大學(xué)理學(xué)院;
【基金】：國家自然科學(xué)基金項目(31401281) 湖南省自然科學(xué)基金項目(14JJ3115) 湖南省高校科技創(chuàng)新團隊支持計劃(2014207) 湖南省科技計劃重點研發(fā)項目(2016NK2151)
【分類號】：TS225.1;O657.3
，

本文編號：2243759