期刊论文

Fingerprints Study on Different Originals'Lentinus edodes by Coupling GC-MS with Random Forest Method

中文

分析科学学报

1006-6144

2019

35

1

85-89

10.13526/j.issn.1006-6144.2019.01.016

中央本级重大增减支项目“名贵中药资源可持续利用能力建设”(2060302).

本校为第一机构

本文采用气相色谱-质谱(GC-MS)方法,并结合随机森林算法对来自不同产地的香菇进行归属分析。首先利用GC-MS法得到各产地香菇的指纹图谱,进一步采用自适应迭代惩罚的偏最小二乘法(adaptive iteratively reweighted Penalized Least Squares,airPLS)和多尺度色谱峰校正,分别对指纹图谱进行背景扣除和保留时间校正等预处理后,再进行定性定量分析,得到各指纹图谱的主要的特征物质,进而采用随机森林算法对不同产地的香菇进行判别分析。本方法将GC-MS法和随机森林算法相融合,能够准确地区分不同产地的香菇样品。

In this study,GC-MS and random forest method were used to analyze and distinguish Lentinus edodes samples collected from different originals.Firstly,the fingerprints of samples were obtained by using GC-MS,then,adaptive iteratively reweighted Penalized Least Squares(airPLS)and multi-scale chromatographic calibration approaches were adopted to process these chromatographic fingerprints,the main characteristic components from these sample were qualitatively and quantitatively analyzed.Based on these obtained characteristic components,random fores...