作　　者： ; ; ; ; ; ;

机构地区： 河南工业大学粮油食品学院

出　　处： 《分析化学》 2013年第9期1366-1372,共7页

摘　　要： 结合粒子群最小二乘支持向量机(PSO-LSSVM)与偏最小二乘法(PLS)提出一种基于气相色谱技术的新方法,对芝麻油进行真伪鉴别,并对掺伪品中掺假比例进行定量分析。采用主成分分析法(PCA)对857个样本的脂肪酸色谱数据进行分析,优选主成分作为最小二乘支持向量机(LSSVM)的输入向量。利用粒子群算法(PSO)优化LSSVM,构建芝麻油掺伪鉴别的两级分类模型,同时运用PLS建立掺伪芝麻油中掺伪油脂的定量校正模型,两级分类模型的准确率分别达到了100%和98.7%,定量分析模型的平均预测标准偏差(RMSEP)为3.91%。结果表明,本方法的鉴别准确性和模型泛化能力均优于经典的BP神经网络和支持向量机(SVM),可用于食用油脂加工和流通环节的质量控制,为食用油质量的准确鉴定提供了一条有效途径。 Through particle swarm optimization（PSO）,least squares support vector machine（LSSVM） and partial least squares（PLS） regression,this study was performed to the development of a new method for detection and quantification of adulteration of sesame oil with vegetable oils using gas chromatographic（GC） technique.Based on principal component analysis（PCA）,the GC data of total 857 samples including 117 authentic sesame oils and 740 adulterated sesame oils were firstly analyzed for dimension reduction.Using the PCA-filtered GC data,a hierarchical approach including two steps was established for the detection and the quantification of oil samples.At the first step,a model was constructed to discriminate between authentic sesame oils and adulterated sesame oils using least squares support vector machine（LSSVM） algorithm.Then,another LSSVM-based model was developed to identify the type of adulterant in the mixed oil.At last,the PLS models were built to quantification of the adulterated oils.The prediction results showed that the classification model could achieve correct rate 100.0% and 98.7%,and the root-mean-square errors of PLS model were 3.91%,meaning that this approach is a valuable tool to detect and quantify the adulteration of sesame oil compared with other methods such as BP neural network and support vector machine.

关 键 词： 芝麻油 最小二乘支持向量机 粒子群优化算法 偏最小二乘法 掺伪

领　　域： [轻工技术与工程] [轻工技术与工程]