人工神经网络用于直接化学电离质谱分析食用油品质的研究

Surface Desorption Atmospheric Pressure Chemical Ionization Mass Spectrometry for Edible Oil Analysis Based on Back Propagation Neural Networks

无需任何样品预处理,采用表面解吸常压化学电离质谱(DAPCI-MS)技术直接对涂覆在载玻片表面的食用油样品和地沟油样品进行检测,快速获得了不同油类样品的质谱信号;并运用改进的反向传输(BP)人工神经网络对DAPCI-MS所得到的油类样品质谱数据进行有监督的分类识别,建立多分组预测模型。结果表明:DAPCI-MS能够承受食用油中复杂基体的影响,可对油类样品进行直接快速质谱分析;误差反转(BP)神经网络具有良好的分类判别能力,对食用油样品质谱数据识别效果比较理想,能够在对地沟油和非地沟油样品进行有效区分的同时,实现对不同品种的食用油的分离及分类判别。本方法分析速度快,信息提取准确,识别精度高,对快速质谱技术结合神经网络在该领域的应用以及食用油品质的快速鉴定具有重要的借鉴意义。

Without any sample pretreatment, oil samples smeared on slide were directly detected by surface desorption atmospheric pressure chemical ionization mass spectrometry （DAPCI-MS）, the mass spectra rapidly recorded by DAPCI-MS were subjected to data processing for classification using improved BP（Back Propagation） neural networks. The results showed that DAPCI-MS was a practically convenient tool for edible and hogwash oil detection without much matrix effect. The improved BP neural network can be applied to the simultaneous determination of hogwash-standard oil and different kinds of edible oil samples. The data demonstrated that the DAPCI-MS combined improved BP neural network methods was a promising technique for edible oil rapid identification with expedite convergence pace and superior prediction precision.