本文已被:浏览 39次 下载 10次
投稿时间:2023-04-03
投稿时间:2023-04-03
中文摘要: 该文采用傅里叶变换中红外光谱和矿质元素信息结合多种特征变量选取方法,建立偏最小二乘判别分析(partial least squares discriminant analysis,PLS-DA)模型实现冰糖橙的产地鉴别,并通过低级和中级两种数据融合策略进一步提高鉴别能力。结果显示,冰糖橙的中红外吸收峰主要与糖类和蛋白质有关,不同产地冰糖橙的中红外光谱在糖类相关的吸收峰上存在差异。不同产地冰糖橙的矿质元素含量有所不同,其中Fe、K、Mg、P、Se 5 种矿质元素在4 个产地的冰糖橙之间呈现显著性差异。采用变量投影重要性指标、反向区间偏最小二乘法和组合区间偏最小二乘法等特征变量选取方法可以提高产地鉴别模型的正确率。与使用单一技术获得数据建立的PLS-DA 模型相比,使用中级数据融合策略建立的模型在训练集和测试集上分别取得98%和100%的正确率。研究表明中红外光谱和矿质元素信息都能反映不同产地冰糖橙间的差异,基于中级数据融合策略融合两种信息建立的PLS-DA 模型能够实现冰糖橙产地的准确鉴别,可以作为冰糖橙溯源研究的一种可靠方法。
中文关键词: 冰糖橙 产地鉴别 傅里叶变换中红外光谱 矿质元素 数据融合
Abstract:In this paper,the partial least squares discriminant(PLS-DA)model was established by using Fourier transform mid-infrared spectra and mineral element information combined with multiple feature selection methods to identify the origin of Bingtang sweet oranges. Low-level and mid-level data fusion strategies were used to further improve the identification performance. The results showed that the mid-infrared spectra of Bingtang sweet oranges were mainly composed of the absorption peaks of sugars and proteins.There were differences mainly in the absorption peaks related to sugars of the mid-infrared spectra among the samples from different origins. The contents of mineral elements of Bingtang sweet orange from different origins were different,among which Fe,K,Mg,P,and Se were significantly different among the four areas. Variable importance in projection(VIP),backward interval PLS(Bipls),and synergy interval PLS(Sipls)were used to improve the accuracy of the origin identification model. Compared with the model established by data obtained by using a single technology,the PLS-DA model established using the mid-level data fusion strategy achieved 98% and 100% accuracy on the training set and the test set. The study showed that both the mid-infrared spectra and mineral element information could reflect the differences of Bingtang sweet oranges from different origins. The PLS-DA model based on a mid-level data fusion strategy can achieve accurate origin identification of Bingtang sweet oranges,which could be used as a reliable method for distinguishing the origin of Bingtang sweet oranges.
keywords: Bingtang sweet oranges origin identification Fourier transform mid-infrared spectra mineral elements data fusion
文章编号:202423018 中图分类号: 文献标志码:
基金项目:国家自然科学基金资助项目(31760341)
作者 | 单位 |
吴衷宇1,汪禄祥2,刘兴勇2,欧全宏1,时有明3,刘刚1* | 1.云南师范大学物理与电子信息学院,云南 昆明 650500;2.云南省农业科学院质量标准与检测技术研究所,云南 昆明 650205;3.曲靖师范学院物理与电子工程学院,云南 曲靖 655011 |
引用文本: