分析测试学报2025，Vol.44Issue(6)：1154-1160,7.DOI:10.12452/j.fxcsxb.241118535
结合近红外光谱和模型更新的苹果品质无损检测
Combining Near-infrared Spectroscopy and Model Updating for Non-destructive Testing of Apple Quality
摘要
Abstract
Varietal differences significantly affect the soluble solid content(SSC)and near-infrared spectroscopy(NIRS)characteristics of apples,creating challenges when applying SSC spectral cali-bration models developed for one variety to others.This study developed a partial least squares regres-sion(PLSR)calibration model using Aksu Fuji apples(Batch 1)and addressed the practical chal-lenge of predicting SSC in Qingdao Scarlet apples(Batch 2)through model updating methods.The PLSR model,created with a combination of first derivative(1D)preprocessing and competitive adap-tive reweighted sampling(CARS),effectively predicted SSC for Batch 1,achieving a correlation co-efficient of prediction(Rp)of 0.972 8 and a root mean square error of prediction(RMSEP)of 0.383 8 °Brix.However,the Batch 1 model performed poorly in predicting SSC for Batch 2.To address this limitation,three model updating methods—calibration updating,slope/bias correction(SBC),and dynamic orthogonal projection(DOP)—were applied,and the impact of different update sample sizes was evaluated.Results showed that RMSEP significantly decreased after model updating.Among the methods,SBC performed best,reducing the RMSEP for Batch 2 from 1.075 6 °Brix to 0.233 4 ° Brix with 20 new samples.These findings demonstrate that model updating effectively improves pre-diction performance across different apple varieties,enhancing model robustness and offering valu-able guidance for maintaining and updating SSC detection models in practical applications.关键词
苹果/可溶性固形物含量/近红外光谱/PLSR模型/模型更新Key words
apple/soluble solids content/near-infrared spectroscopy/PLSR model/model up-date分类
化学引用本文复制引用
吴琪,陈孝敬,石文,谢忠好,苏来金,黄光造..结合近红外光谱和模型更新的苹果品质无损检测[J].分析测试学报,2025,44(6):1154-1160,7.基金项目
国家自然科学基金青年项目(62305253) （62305253）
