蚁群-遗传算法优化近红外光谱检测花茶花青素 含量的研究

Application of ACO-iPLS and GA-iPLS for wavelength selection in NIR spectroscopy

中文关键词:  花茶  花青素  蚁群-遗传算法  近红外光谱  定量分析模型

英文关键词:scented tea  anthocyanin  ACO-GA-iPLS  near infrared spectroscopy  quantitative analysis model

中文摘要:

      目的 本研究基于蚁群-遗传区间偏最小二乘(ACO-GA-iPLS)近红外谱区筛选方法预测花茶花青素含量。方法 首先对花茶近红外光谱进行预处理; 然后用ACO-iPLS优选出特征子区间; 最后对所选的特征子区间, 用GA-iPLS进一步细化花青素的特征子区间, 并建立花青素的预测模型。结果 优选出3个特征子区间(第1、9、10子区间), 所建模型对应的交互验证均方根误差(RMSECV)和预测均方根误差(RMSEP)分别为0.1460 mg/g和0.1840 mg/g, 校正集和预测集相关系数分别为0.9187和0.8856。结论 ACO-GA-iPLS可以有效选择近红外光谱特征波长, 简化模型, 提高模型精度。

英文摘要:

      Objective In order to improve the prediction accuracy of quantitative analysis model of NIR spectroscopy, this study proposed a method to select the optimal spectra intervals from the whole NIR spec-troscopy, and predict the anthocyanin content of scented tea. Methods Raw NIR spectra of scented tea samples were preprocessed by SNV, then wavelength regions were selected by ant colony optimization (ACO) algorithm. Finally, the genetic algorithm-interval partial least squares was used to refine the wavelength regions selected by ACO, and predict the anthocyanin content of scented tea. Results The scented tea spectra were divided into 12 intervals, among which 3 subsets, i.e. No. 1, 9, 10 were selected by ACO-iPLS. Then, the selected wavelength regions set were divided into 12 intervals and selected by GA-iPLS. The optimal iPLS model was built with the RMSECV and RMSEP were 0.1460 mg/g and 0.1840 mg/g, and the calibration and prediction correlation coefficient were 0.9187 and 0.8856, respectively. Conclusion The ACO-GA-iPLS can effectively select wavelength regions from near infrared spectroscopy, simplify model complexity and improve accurately of model.