基于嗅觉可视化技术的花生黄曲霉毒素B_(1)定量检测

Quantitative Determination of Aflatoxin B_(1)(AFB_(1))in Peanuts Based on Olfaction Visualization Technology

花生在储运过程中极易受到各种霉菌的污染而产生真菌毒素,其中以黄曲霉毒素B_(1)(AFB_(1))最为常见。提出了基于嗅觉可视化技术的花生AFB1定量检测。利用顶空固相微萃取气相色谱-质谱联用技术(HS-SPEM-GC-MS)分析得到不同霉变花生的指示性挥发性物质成分,据此选择12种化学染料制备特异性强的色敏传感器阵列,用于采集不同霉变程度花生样本的气味信息。引入遗传算法(GA)结合反向传播神经网络(BPNN)优化预处理后的传感器特征图像的颜色分量。借助支持向量回归(SVR)构建基于优化特征颜色分量组合的定量模型实现花生AFB1的定量检测;在此过程中,比较网格搜索(GS)和麻雀搜索算法(SSA)对SVR参数的优化性能。研究结果显示:SSA-SVR模型性能整体优于GS-SVR模型性能;且基于7个特征颜色分量组合的最佳SSA-SVR模型的预测相关系数(RP)达到0.9142,预测均方根误差为5.6832μg/kg,剩余预测偏差为2.3926。研究结果表明,利用嗅觉可视化技术可实现花生AFB1的定量检测。

Peanuts are easily contaminated by various molds during storage and transportation to produce mycotoxins,among which aflatoxin B_(1)(AFB_(1))is the most common.A novel method for determination of the AFB1in peanuts based on colorimetric senor technology was proposed.Indicative volatile components of different moldy peanuts were obtained by headspace solid phase microextraction with gas chromatography-mass spectrometry(HS-SPEM-GC-MS).According to the result of the HS-SPEM-GC-MS report,totally 12 kinds of chemical dyes were selected to prepare a color sensitive sensor array with strong specificity,which was used to collect the odor information of peanut samples with different degrees of mildew.Genetic algorithm(GA)combined with back propagation neural network(BPNN)was used to optimize the color component of the preprocessed sensor feature image.Then support vector regression(SVR)was used to construct a quantitative model based on the combination of optimized feature color components to achieve the determination of the AFB1in peanuts.In this process,the optimization performance of grid search(GS)algorithm and sparrow search algorithm(SSA)on SVR parameter was compared.The results obtained showed that the performance of SSA-SVR model was better than that of GS-SVR model.The correlation coefficients of prediction(RP)of the best SSA-SVR model based on the combination of seven feature color components reached 0.9142.The root mean square error of prediction(RMSEP)was 5.6832μg/kg,and the residual predictive deviation(RPD)value was 2.3926.The overall results demonstrated that it was feasible to use the olfactory visualization technology for the determination of the AFB1in peanuts.In addition,proper optimization of sensor features and model parameters can further improve the detection performance of the model.