摘要
提出了利用近红外光谱技术对重金属污染泥蚶的快速检测方法。以冷冻干燥磨粉的泥蚶肌肉为试验对象,方案设计分为两种:对照泥蚶和单一重金属(Cd,Cu,Pb或Zn)污染泥蚶的分类(设计Ⅰ);所有样本包括对照泥蚶和4种重金属污染泥蚶的分类(设计Ⅱ)。采用两种识别算法,即最小二乘支持向量机和随机森林,对设计I和设计II分别建立分类模型并进行预测。预测结果表明:设计Ⅰ:最小二乘支持向量机和随机森林的平均预测正确率分别为100%和95%;设计Ⅱ:最小二乘支持向量机和随机森林的平均预测正确率分别为96%和92%。利用近红外光谱技术快速检测重金属污染泥蚶具有可行性,可为泥蚶重金属污染提供一种快速检测方法。
This study proposed the rapid detection of heavy metal-contaminated Tegillarca granosa by using near infrared spectroscopy.In this paper, based on the freeze-dried and grounded Tegillarca granosa, the experimental design was divided into： classify Tegillarca granosa samples that were uncontaminated（control） and contaminated by a certain heavy metal（Cd, Cu, Pb, or Zn）（Design I）; classify all sample varieties, including the samples that were uncontaminated and contaminated by the four heavy metals（Design II）. Two recognition algorithms, namely, the least-squares support vector machines（LS-SVM） and random forest（RF） were used to construct classification models and prediction for design I and II, respectively. Prediction results displayed that the average prediction accuracy of LS-SVM and RFreached100% and 95% for Design I respectively, and reached 96% and 92% for Design II respectively. The results of this study indicated the potential of near infrared spectroscopy in evaluating heavymetal contamination in Tegillarca granosa. Meanwhile, it could provide a rapid detection method for heavy metal contamination in Tegillarca granosa.
出处
《中国食品学报》
EI
CAS
CSCD
北大核心
2015年第4期189-195,共7页
Journal of Chinese Institute Of Food Science and Technology
基金
国家自然科学基金项目(31201355)
关键词
重金属
近红外光谱
水产品
重金属污染
泥蚶
最小二乘支持向量机
随机森林
heavy metal
near infrared spectroscopy
aquatic product
heavy metal contamination
Tegillarca gra-nosa
least-squares support vector machines
random forest
