提出了一种应用近红外光谱技术快速无损鉴别橄榄油产地的新方法。采用近红外光谱仪获取三种不同产地的橄榄油各30个样本的光谱漫反射特征曲线,利用全局搜索算法-遗传算法提取特征波长,即从光谱751个波长数据提取9个特征波长数据,并将其作为主成分分析法的输入变量,运用主成分分析法建立分析校正模型。结果表明,主成分1和2累计可信度已达99.130%,对不同产地的橄榄油有很好的聚类作用,同时也说明遗传算法抽取特征波长方法正确。将提取到的六种主成分作为BP神经网络的输入变量,品种类型作为神经网络的输出变量,建立3层人工神经网络模型,对30个未知橄榄油产地进行预测,预测结果准确率达100％。该方法能快速无损地检测橄榄油产地,同时也为其他油类产地鉴别提供了一种新方法。

A new method for the fast discrimination of different producing areas of olive oil by means of near infrared spectroscopy (NIRS) was developed.A relation was established between the reflection spectra and three varieties of olive oil from different places.The data set of modeling consists of a total of 90 samples of olive oil and each type consists of 30 samples.Genetic algorithms (GA),a global searching method,was applied to select the key features of the wavelengths.By the treatment with GA,the quantitative information was obtained and the number of characteristics for principal component analysis (PCA) was reduced to 9.By the treatment with PCA,the quantitative information was obtained and the number of characteristics for BP (back propagation) neural network was reduced to 6.The analysis suggests that the cumulate reliabilities of PC1 and PC2 (the first two principal components) are higher than 99%.It appeared to provide the best clustering of the different areas of olive oil and the results show that it is successful to use the GA to extract the key features of spectral wavelengths of olive oil.The first 6 principal components were used for modeling parameters of BP neural network model and the area sorts of olive oil were used for parameters of export.Three layers of neural network model were built up to predict the 30 unknown samples.The recognition rate of 100% was achieved.It can be concluded that the method is quite suitable for the fast discrimination of producing areas of olive oil and also offers a new-approach to the discrimination of producing areas of other oils.