近红外光谱结合BP神经网络法快速分析薰衣草精油中主要成分
|
摘要
[目的]建立薰衣草精油主体成分的快速测定模型,以满足中国薰衣草精油质量标准体系中对化学成分分析的要求。[方法]采用近红外光谱方法结合BP神经网络快速测定薰衣草精油中芳樟醇和乙酸芳樟酯的含量,在7 000～4 000 cm~(-1)波数范围内,薰衣草精油化学信息量比较丰富且噪音低,可作为分析区间。通过间隔偏最小二乘方法(iPLS)选择与测定指标相关性高的近红外波长段集合,分别得到乙酸芳樟酯和芳樟醇的185个定量分析用波长点数据集,对BP神经网络参数优化后,建立新疆薰衣草精油中芳樟醇和乙酸芳樟酯的iPLS-BP神经网络快速近红外定量分析模型。[结果]采用方法的芳樟醇和乙酸芳樟酯测定系数分别为0.972 91和0.946 90,预测结果与GC-MS分析的相对误差分别小于2.45%和2.85%。[结论]定量分析模型能够很好地快速测定薰衣草精油中乙酸芳樟酯、芳樟醇含量,具有良好的预测能力,可为新疆薰衣草精油主要成分的快速定量分析提供一种新的有效方法。
[Objective] The research aimed to establish a rapid determination model of lavender essential oil body composition, it can meet the requirements of chemical components in the quality standard system of Chinese lavender essential oil. [Method]We can quickly determine the content of linalool and linalyl acetate in the essential oil of Xinjiang lavender by using near infrared spectra combining BP neural network. On the process of this study measured the lavender essential oil samples of near infrared absorption spectrum, by analysis of near infrared spectral absorption peak by the number of wavelengths, 7 000-4 000 cm~(-1) within the scope of lavender essential oil chemical information is abundant in Xinjiang and the wavelength section of low noise, so choose the wave number range for analysis. We obtained 185 data sets of wavelength points for quantitative analysis of linalool and linalyl acetate respectively, through interval partial least squares method(iPLS) selection high correlation of the near infrared wavelength section. After optimizing the parameters of BP neural network,the model of fast near infrared quantitative by iPLS-BP neural network method is established. [Result]The measurement coefficient of linalool and linalyl acetate by iPLS-ANN method was 0.972 91 and 0.946 90 respectively, and the relative error between the prediction and GC-MS analysis was less than 2.45% and 2.85%.[Conclusion]The quantitative analysis model can quickly determine the content of linalool and linalyl acetate in lavender essential oil, and has good predictive ability, which can provide a new effective method for the rapid quantitative analysis of main components of lavender essential oil in Xinjiang.
[Objective] The research aimed to establish a rapid determination model of lavender essential oil body composition, it can meet the requirements of chemical components in the quality standard system of Chinese lavender essential oil. [Method]We can quickly determine the content of linalool and linalyl acetate in the essential oil of Xinjiang lavender by using near infrared spectra combining BP neural network. On the process of this study measured the lavender essential oil samples of near infrared absorption spectrum, by analysis of near infrared spectral absorption peak by the number of wavelengths, 7 000-4 000 cm~(-1) within the scope of lavender essential oil chemical information is abundant in Xinjiang and the wavelength section of low noise, so choose the wave number range for analysis. We obtained 185 data sets of wavelength points for quantitative analysis of linalool and linalyl acetate respectively, through interval partial least squares method(iPLS) selection high correlation of the near infrared wavelength section. After optimizing the parameters of BP neural network,the model of fast near infrared quantitative by iPLS-BP neural network method is established. [Result]The measurement coefficient of linalool and linalyl acetate by iPLS-ANN method was 0.972 91 and 0.946 90 respectively, and the relative error between the prediction and GC-MS analysis was less than 2.45% and 2.85%.[Conclusion]The quantitative analysis model can quickly determine the content of linalool and linalyl acetate in lavender essential oil, and has good predictive ability, which can provide a new effective method for the rapid quantitative analysis of main components of lavender essential oil in Xinjiang.
引文
[1] CAVANAGH H M A,WILKINSON J M.Biological activities of lavender essential oil [J]. Phytother Res, 2002,16(4):301-308.
[2] 徐洁华,文首文. 薰衣草挥发性有机物及其药理功效研究进展[J].时珍国医国药,2010,21(4):979-980,983.
[3] 国家卫生和计划生育委员会.食品添加剂薰衣草油:GB 1886.38—2015[S].北京:中国标准出版社,2016.
[4] 王菊香,邢志娜,李伟,等. BP神经网络法结合红外光谱快速测定在用润滑油胺类抗氧剂含量[J].计算机与应用化学, 2016,33(2):197-199.
[5] 韩晓,王菊香,刘洁.基于BP-神经网络的航空煤油总酸值近红外光谱快速检测[J].分析科学学报, 2011, 27(6):751-754.
[6] 李智,王圣毫,郑维平,等.基于傅立叶变换的人工神经网络近红外光谱定量分析法[J].分析测试学报,2012,31(3):343-346.
[7] 王书涛,陈东营,魏蒙,等.荧光光谱法和PSO-BP神经网络在山梨酸钾浓度检测中的应用[J].中国激光,2015,42(5):304-310.
[8] 邹小波,朱曾,赵杰文.基于间隔偏最小二乘法的农产品近红外光谱谱区选择方法[J].现代科学仪器,2007(1):86-88.
[9] 迟亮,张贺龙,车英,等. 基于iPLS的玉米近红外光谱数据处理方法研究[J]. 长春理工大学学报(自然科学版),2011,34(2):117-119.
[10] 李强,杨天邦,涂公平.GA-BP神经网络模型应用于岩芯扫描仪测定海洋沉积物中多种组分的半定量分析[J].分析仪器, 2018(1):75-79.
[11] 曹家兴,陆建平.遗传算法-贝叶斯正则化BP神经网络拟合滴定糖蜜中有机酸[J].分析化学,2011,39(5):743-747.
[12] 徐立鹏,葛良全,谷懿,等.基于PCA-BP神经网络的EDXRF分析测定地质样品中铁、钛元素含量的应用研究[J].光谱学与光谱分析,2013,33(5):1392-1396.
[13] 杰尔·沃克曼,洛伊斯·文依.近红外光谱解析实用指南[M].褚小立,许育鹏,田高友,译.北京:化学工业出版社,2009.
[2] 徐洁华,文首文. 薰衣草挥发性有机物及其药理功效研究进展[J].时珍国医国药,2010,21(4):979-980,983.
[3] 国家卫生和计划生育委员会.食品添加剂薰衣草油:GB 1886.38—2015[S].北京:中国标准出版社,2016.
[4] 王菊香,邢志娜,李伟,等. BP神经网络法结合红外光谱快速测定在用润滑油胺类抗氧剂含量[J].计算机与应用化学, 2016,33(2):197-199.
[5] 韩晓,王菊香,刘洁.基于BP-神经网络的航空煤油总酸值近红外光谱快速检测[J].分析科学学报, 2011, 27(6):751-754.
[6] 李智,王圣毫,郑维平,等.基于傅立叶变换的人工神经网络近红外光谱定量分析法[J].分析测试学报,2012,31(3):343-346.
[7] 王书涛,陈东营,魏蒙,等.荧光光谱法和PSO-BP神经网络在山梨酸钾浓度检测中的应用[J].中国激光,2015,42(5):304-310.
[8] 邹小波,朱曾,赵杰文.基于间隔偏最小二乘法的农产品近红外光谱谱区选择方法[J].现代科学仪器,2007(1):86-88.
[9] 迟亮,张贺龙,车英,等. 基于iPLS的玉米近红外光谱数据处理方法研究[J]. 长春理工大学学报(自然科学版),2011,34(2):117-119.
[10] 李强,杨天邦,涂公平.GA-BP神经网络模型应用于岩芯扫描仪测定海洋沉积物中多种组分的半定量分析[J].分析仪器, 2018(1):75-79.
[11] 曹家兴,陆建平.遗传算法-贝叶斯正则化BP神经网络拟合滴定糖蜜中有机酸[J].分析化学,2011,39(5):743-747.
[12] 徐立鹏,葛良全,谷懿,等.基于PCA-BP神经网络的EDXRF分析测定地质样品中铁、钛元素含量的应用研究[J].光谱学与光谱分析,2013,33(5):1392-1396.
[13] 杰尔·沃克曼,洛伊斯·文依.近红外光谱解析实用指南[M].褚小立,许育鹏,田高友,译.北京:化学工业出版社,2009.