基于拉曼光谱和最小二乘支持向量机的橄榄油掺伪检测方法研究

为实现橄榄油中掺伪油类型的识别和掺伪量预测,对掺入葵花籽油、大豆油、玉米油的橄榄油共117个样品进行拉曼光谱检测,并用基于多重迭代优化的最小二乘支持向量机模型对掺入油的类型进行识别,综合识别率为97%。同时分别采用最小二乘支持向量机、人工神经网络模型、偏最小二乘回归建立橄榄油中葵花籽油、大豆油、玉米油含量的拉曼光谱定标模型,结果显示最小二乘支持向量机具有最优的预测效果,其预测均方根误差(RMSEP)在0.007 4～0.014 2之间。拉曼光谱结合最小二乘支持向量机可为橄榄油掺伪检测提供一种精确、快速、简便、无损的方法。

Research on detection method of adulterated olive oil by Raman spectroscopy and least squares support vector machine

For the purpose of the authentication of sorts as well as the prediction of contents of the oils which were adulterated into olive oil, 117 olive oil samples adulterated with sunflower seed oil, soybean oil and corn oil were detected by Raman spectroscopy, and least squares support vector machine (LS-SVM) based on multiple iterative optimization was used to identify the type of the adulterant oil, and the composite recognition rate was 97%. In addition, methods such as LS-SVM, ANNs and PLSR were used to build the Raman spectra calibration model of the adulterant oil (sunflower seed oil, soybean oil and corn oil) contents respectively, the results indicated that LS-SVM had the best predictive performance, and the root mean square error of prediction (RMSEP) ranged from 0.007 4 to 0.014 2. Research results showed the method based on Raman spectroscopy and LS-SVM was accurate, fast, simple and non-destructive for adulterated olive oil detection.