分析测试学报2025，Vol.44Issue(6)：1086-1095,10.DOI:10.12452/j.fxcsxb.25012462
基于FT-NIR技术结合化学计量学方法快速准确鉴别天麻不同栽培品种
Rapid and Accurate Identification of Gastrodia elata Blum Cultivar Based on FT-NIR Spectroscopy Combined with Chemometric Methods
摘要
Abstract
In this study,a partial least squares discriminant analysis(PLS-DA)model and a residu-al convolution neural network(ResNet)model were constructed using Fourier transform near-infrared spectroscopy(FT-NIR)and two-dimensional correlation spectroscopy(2DCOS)technology,in con-junction with chemometric methods and deep learning algorithms,to rapidly and accurately identify three cultivated varieties of Gastrodia elata Blum(G.elata Bl.)samples(447).The results showed that the PLS-DA model,created by integrating first derivativ(1st Der)and multiple scatter correc-tion(MSC)preprocessing of FT-NIR data,demonstrated the highest stability and the best overall performance,with an accuracy of 99.00%.At the same time,the identification method based on FT-NIR synchronous 2DCOS image combined with ResNet model could achieve rapid and accurate identification(100.00%accuracy)of different cultivars of G.elata Bl without the need for optimal pretreatment and complex data conversion.This study provides a rapid and accurate method for iden-tifying different cultivars of G.elata Bl.,and lays a foundation for further germplasm resource re-search and breeding of new variety.关键词
傅里叶变换近红外光谱/化学计量学/机器学习/天麻/栽培品种Key words
Fourier transform near-infrared spectroscopy/chemometrics/machine learning/Gas-trodia elata Blum/cultivar分类
化学化工引用本文复制引用
苏俊宇,刘鸿高,王元忠..基于FT-NIR技术结合化学计量学方法快速准确鉴别天麻不同栽培品种[J].分析测试学报,2025,44(6):1086-1095,10.基金项目
国家自然科学基金项目(82460746) （82460746）
