文章摘要
李艳芬,马瑞峻,陈瑜,黄丽,颜振锋,蔡祥.光谱技术结合化学计量学分析方法快速检测阿维菌素的试验研究[J].农业环境科学学报,2023,42(9):2140-2146.
光谱技术结合化学计量学分析方法快速检测阿维菌素的试验研究
An experimental study on rapid detection of the biopesticide avermectin using spectroscopy combined with stoichiometric analysis
投稿时间:2022-12-14  修订日期:2023-03-03
DOI:10.11654/jaes.2022-1269
中文关键词: 阿维菌素  紫外/可见吸收光谱  化学计量学方法  快速检测  定量分析
英文关键词: avermectin  ultraviolet/visible absorption spectra  stoichiometry methods  rapid detection  quantitative analysis
基金项目:国家重点研发计划项目(2016YFD0800901)
作者单位E-mail
李艳芬 华南农业大学工程学院, 广州 510642  
马瑞峻 华南农业大学工程学院, 广州 510642 maruijun_mrj@163.com 
陈瑜 华南农业大学工程学院, 广州 510642 chenyu219@126.com 
黄丽 华南农业大学工程学院, 广州 510642  
颜振锋 华南农业大学工程学院, 广州 510642  
蔡祥 华南农业大学工程学院, 广州 510642  
摘要点击次数: 960
全文下载次数: 827
中文摘要:
      生物农药阿维菌素是一种微溶于水的高毒杀虫剂,有广谱、高效等特点,在防治水稻螟虫,稻纵卷叶螟方面表现优异,但对水生生物高毒。为了探讨利用光谱技术现场快速检测水体中生物农药阿维菌素的可能性,测定了其在紫外/可见光波长范围内的不同浓度吸光度光谱数据,建立其快速有效的定量分析模型。使用不同光程比色皿采集一定浓度范围的阿维菌素农药样本光谱数据进行对比,得到最佳光谱数据用于后续定量处理分析。将波长范围为200~500 nm的光谱数据采用Savitzky-Golay卷积平滑法(S-G平滑法)进行数据预处理,将原始光谱数据和S-G平滑法预处理后的光谱数据校正集和预测集分别按不同比例采用Sample set partitioning based on joint x-y distance(SPXY)算法进行样本集划分,并分别建立PLS模型进行比较。再将划分样本集后优选出的光谱数据采用主成分分析(Principle component analysis,PCA)结合马氏距离阈值法(Mahalanobis Distance,MD),即PCAMD算法剔除异常样本,再将剔除异常样本后的光谱数据采用竞争性自适应重加权采样法(Competitive adaptive reweightedsampling,CARS)筛选特征波长变量,建立S-G平滑-SPXY-(PCA-MD)-CARS-PLS定量分析模型。结果表明,100 mm光程比色皿获得的光谱数据最佳,245.4 nm处为阿维菌素特征吸收峰。原始数据经S-G平滑法预处理、SPXY划分样本集、PCA-MD剔除异常样本以及CARS筛选特征波长变量后建立的定量模型最优,模型评价系数R2p为0.998 8,预测集均方根误差为0.061 1,剩余预测残差为29.589 4,该方法有效简化了模型并提高了模型精度和稳健性。研究表明,生物农药阿维菌素的紫外/可见光吸收光谱数据结合化学计量学分析方法能够用来定量分析生物农药阿维菌素浓度。
英文摘要:
      The biopesticide abamectin is a highly toxic insecticide that is slightly soluble in water. It exhibits a broad spectrum of application and high efficiency. Abamectin shows excellent performance in controlling rice borers and rice leaf rollers, however, it is highly toxic to aquatic organisms. To test the feasibility of rapid detection of avermectin in water through spectroscopy, absorption spectral data of different concentrations in the ultraviolet/visible wavelength range were measured, and a rapid and effective quantitative analysis model was established. Spectroscopic data of avermectin pesticide samples in a certain concentration range were collected with different optical path cuvettes for comparison, and the best spectral data were used for subsequent quantitative processing and analyses. Spectral data with a wavelength range of 200-500 nm were preprocessed using the Savitzky-Golay convolutional smoothing(S-G smoothing)method, and original spectral data and spectral data preprocessed through S-G smoothing were used to divide the samples set according to different proportions using sample set partitioning based on a joint x-y distance(SPXY)algorithm, and PLS models for comparison were established. Then, the spectral data selected after dividing the sample set were selected through principal component analysis(PCA)combined with a Mahalanobis Distance(MD)(PCA-MD)algorithm to eliminate abnormal samples. After removing abnormal samples, the spectral data were screened using the competitive adaptive reweighted sampling(CARS)method to screen the characteristic wavelength variables, and the quantitative analysis model of S-G smoothing-SPXY-(PCA-MD)-CARS-PLS was established. Spectral data results obtained from 100 mm optical path cuvettes were the best, and the characteristic absorption peak of avermectin was at 245.4 nm. The quantitative model established after S-G smoothing preprocessing, SPXY dividing of the sample set, removing abnormal samples by PCA-MD, and screening characteristic wavelength variables by CARS was optimal; the model evaluation coefficient R2p was 0.998 8, RMSEP was 0.061 1, and RPD was 29.589 4. The model was effectively simplified, and its accuracy and robustness were improved. The UV/Vis absorption spectral data of the biopesticide avermectin combined with chemometric analysis can be used to quantify avermectin concentrations.
HTML    查看全文   查看/发表评论  下载PDF阅读器