中文摘要：

对常见食用油和地沟油进行了基于激光诱导击穿光谱鉴别研究,建立了人工神经网络模型,预测检验结果良好.选择本地超市常见的2种食用油和大连市产品质量监督检验所提供的地沟油,以定量无尘分析滤纸为基底,获得了滤纸、豆油、调和油、地沟油4种样品各140组LIBS光谱数据,提取了24个特征谱线进行主成分分析,各类样品在主成分空间中呈现良好聚集分类.将得到的140组光谱数据,100组作为训练集,建立人工神经网络模型,40组作为测试数据进行鉴别,识别率达98.1%.基于主成分分析建立的人工神经网络模型识别率达94.2%.研究结果表明,基于主成分分析和人工神经网络的激光诱导击穿光谱检测技术为地沟油快速高效鉴别研究带来了新的思路与方法,对地沟油的鉴别具有十分重要的意义.

英文摘要：

In order to distinguish edible oil and trench oil （which is called di-gou-you in Chinese）, spectral features of the samples were investigated by laser-induced breakdown spectroscopy （LIBS） in this paper. Artificial neural network （ANN） model was also built and tested for spectral data analysis, and the research results showed that the ANN model was reasonably efficient. Two kinds of common edible oil and a kind of typical trench oil were chosen for experimental investigations. Totally, we tested four kinds of samples, including filter paper as sample-matrix for laser ablation, soybean oil, blend oil and trench oil. One hundred and forty LIBS spectra for each sample were collected, and the extracted 24 characteristic spectral lines have been adopted for the principal component analysis （PCA）. The resulting data showed that each kind of sample tied together in principal components space and the four samples were separated well. For ANN statistical analysis, the observed 100 LIBS spectra for each sample were chosen as training set to build ANN model. And the other 40 LIBS spectra of each sample were used as the experimental data analysis. The achieved identification accuracy based on PCA and ANN was 94.2%, 98.1%, respectively. This indicates that LIBS detecting technique based on PCA and ANN statistical analysis brings a new possible approach to identify trench oil efficiently and quickly.