近红外光谱技术定量分析小麦中黄曲霉毒素B1

Quantitative analysis of aflatoxin B1 in Triticum aestivum L. by near-infrared spectroscopy technology

中文关键词:  近红外光谱技术  AFB1  定量分析  无损检测  小麦

英文关键词:near-infrared spectroscopy technology  aflatoxin B1  quantitative analysis  non-destructive testing  Triticum aestivum L.

中文摘要:

      目的 基于近红外光谱技术建立定量预测模型, 实现快速无损测定小麦籽粒中黄曲霉毒素B1 (aflatoxin B1, AFB1)含量。方法 采集小麦样品在900~1700 nm波长范围内的反射光谱, 用高效液相色谱法测定小麦中AFB1含量, 将小麦样品原始光谱数据进行预处理, 提取特征波长, 分别通过反向传播神经网络(back propagation neural network, BPNN)、随机森林(random forest, RF)和支持向量机(support vector machine, SVM)建立AFB1含量预测模型, 并与全波段建模结果进行比较。结果 经多元散射矫正(multiplicative scatter correction, MSC)和竞争性自适应加权算法(competitive adaptive reweighted sampling, CARS)处理后建立的SVM模型优于其他模型和全波段建模模型。结论 结合CARS算法有效提取了AFB1值的特征波长, MSC-CARS-SVM模型能够用于AFB1含量的快速、无损检测, 利用近红外光谱技术实现对AFB1含量的定量分析是可行的, 可通过该方法实现储藏期间小麦品质的检测研究。

英文摘要:

      Objective To achieve rapid and non-destructive determination of aflatoxin B1 (AFB1) content in Triticum aestivum L. kernels by establishing a quantitative prediction model based on near-infrared spectroscopy technology. Methods The reflectance spectra of Triticum aestivum L. samples in the wavelength range of 900–1700 nm were collected, and the AFB1 content in Triticum aestivum L. was determined by high performance liquid chromatography. The raw spectral data of the Triticum aestivum L. samples were subjected to preprocessing, and the feature wavelengths were extracted in order to establish a prediction model. A model for predicting the AFB1 content was developed using a back propagation neural network (BPNN), random forest (RF), and support vector machine (SVM), the results of this model were compared with those of a full-wavelength modelling approach. Results The SVM model constructed following the application of multiplicative scatter correction (MSC) and competitive adaptive reweighted sampling (CARS) processing demonstrates superior performance compared to the other models and the full-band modelling model. Conclusion The combination of the CARS algorithm and the MSC-CARS-SVM model allows for the rapid and non-destructive detection of AFB1 content. The feasibility of using near-infrared spectroscopy for quantitative analysis of AFB1 content has been demonstrated, and this approach can be employed to assess the quality of Triticum aestivum L. during storage.