摘要
为了寻找快速判别山茶油掺假的检测方法,本研究利用近红外光谱技术对掺杂大豆没油山茶油进行掺假检测研究。试验在350~1 800 nm波段范围内采集样本的透射光谱,利用CARS方法筛选重要的波长变量,应用偏最小二乘-线性判别分析（PLS-LDA）建立山茶油掺假的判别模型,并与未经变量优选的判别模型进行比较。结果表明,近红外光谱技术联合CARS-PLS-LDA方法可以有效判别纯山茶油和掺假山茶油,校正集、预测集及独立样本组样本的判别正确率、灵敏度及特异性均为100%。CARSPLS-LDA判别模型性能优于未经变量优选的判别模型,表明CARS方法可以有效筛选重要波长变量,能简化判别模型及提高判别模型的稳定性和判别精度。本研究可为山茶油掺假快速检测提供理论依据。
The quality of camellia oil is excellent and the price is higher than other common edible vegetable oils. In order to seek exorbitant profits, unscrupulous traders will adulterate camellia oils with common edible vegetable oils. So it is important to discriminate camellia oil adulteration quickly and protect benefits of consumers. In order to find a rapid detection method for discrimination of camellia oil adulteration, near infrared spectroscopy was used to discriminate camellia oil adulterated with soybean oil in this research. The objective of this research was to discriminate camellia oil adulterated with soybean oil by near infrared spectroscopy. Transmission spectra of samples were acquired by a QualitySpec spectrometer in the wavelength range of 350 to 1 800 nm. CARS （Competitive adaptive reweighted sampling） method was used to select important wavelength variables, then partial least squares-linear discriminant analysis （PLS-LDA） was conducted to develop discrimination model for camellia oil adulteration, and the discrimination model was compared to other discrimination models without variable selection. The results indicated that near infrared spectroscopy combined with CARS-PLS-LDA can discriminate pure camellia oil from adulterated camellia oil effectively, the accuracy, sensitivity and specificity of samples in calibration, prediction and independent sets are all 100%. Discrimination model of CARS-PLS-LDA is superior to other discrimination models without variable selection, and this indicates that CARS is able to select important wavelength variables effectively. Also it can simplify discrimination model and improve precision and stability of discrimination model. This research can provide theoretical foundation for rapid detection of camellia oil adulteration.
出处
《核农学报》
CAS
CSCD
北大核心
2015年第5期925-931,共7页
Journal of Nuclear Agricultural Sciences
基金
国家自然科学基金(31271612)
江西省自然科学基金(20132BAB214010)
江西省教育厅科学研究基金(GJJ13254)
江西农业大学科学研究基金(QN201105)
