项目名称: 基于近红外中红外数据融合的农药有效成分分析
项目编号: No.31301685
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 农业科学
项目作者: 熊艳梅
作者单位: 中国农业大学
项目金额: 22万元
中文摘要: 农药质量与农业生产、农产品安全和食品安全密切相关。本项目采用数据融合方法,通过谱图融合和特征融合,同时提取农药的近红外中红外吸收信息,建立融合模型。考察不同的光谱预处理方法(平滑、归一化、导数、多元散射矫正等)、特征提取方法(PCA、WT、LDA、UVE等)和多元校正算法(PLS、MLR、PCR、ANN等)对融合模型的影响。选出建立优秀融合模型的光谱预处理方法、特征提取方法和多元校正算法,以建立提取农药的近红外中红外弱吸收信息的有效方法,提高近红外中红外方法测定农药有效成分含量的准确度,应用于低浓度农药分析及农药质量控制。研究对近红外中红外在其它方面的应用具借鉴意义,进一步丰富近红外中红外应用理论。
中文关键词: 农药分析;数据融合;近红外;中红外;
英文摘要: Pesticide quality is closely related to agricultural production and food safety. This project applies spectrum and feature fusion, extracts pesticide information of near infrared (NIR)/mid infrared (MIR) to construct fusion model. The effect of spectra preprocessing method (smooth, normalized, derivative, multiplicative scatter correction, etc.), feature extraction methods (PCA, WT, LDA, UVE, etc.) and multivariate calibration algorithm (PLS, MLR, PCR, ANN, etc.) on fusion model are studied. Establish the effective method of extracting NIR/MIR weak absorption information through selecting the excellent preprocessing methods, feature extraction methods and multivariate calibration algorithm which improve the accuracy of determining pesticide active ingredient by NIR/MIR and apply it in analysis of low concentration pesticide and its quality control. This study is of great significance in the application of NIR/MIR in other aspects and further enriches the application theory of NIR/MIR.
英文关键词: pesticide analysis;data fusion;near-infrared;mid-infrared;