基于可见/近红外光谱西葫芦硬度的无损检测模型的建立  

Establishment of Nondestructive Testing Model of Zucchini Hardness Based on Visible/Near Infrared Spectrum

作　　者：邢书海 张淑娟[1] 孙海霞[1] 陈彩虹 李成吉 XING Shuhai;ZHANG Shujuan;SUN Haixia;CHEN Caihong;LI Chengji(College of Engineering,Shanxi Agricultural University,Taigu 030801,China)

机构地区：[1]山西农业大学工学院

出　　处：《山西农业科学》2020年第1期58-60,121,共4页Journal of Shanxi Agricultural Sciences

基　　金：国家自然科学基金项目（31271973）;晋中市科技重点研发计划项目（Y172007-4）;山西省教育厅支持地方高校人才引进计划项目（J201888011）

摘　　要：利用近红外光谱(350~2500 nm)系统采集180个西葫芦样本的光谱数据,运用多种预处理方法对原始光谱数据进行处理,建立西葫芦果肉硬度的PCR、SMLR和PLSR预测模型;并通过对不同的建模模型进行分析,对西葫芦硬度进行快速检测,实现可见/近红外光谱技术对西葫芦的硬度品质在线无损检测。结果表明,经过卷积平滑法和标准正态变换(S-G+SNV)处理建立的PLSR硬度预测模型效果最好,校正集相关系数为0.979,预测集相关系数为0.976;验证模型结果预测相关系数为0.886,预测均方根误差为0.126。运用可见/近红外光谱技术对西葫芦硬度指标的预测研究具有可行性,研究结果可为今后在线快速无损检测果蔬硬度提供理论依据。The spectral data of 180 zucchini samples were collected by near-infrared spectroscopy(350-2500 nm)system.The original spectral data were processed by various pretreatment methods to establish PCR,SMLR and PLSR prediction models for the hardness of zucchini pulp.The different modeling models were analyzed to test the hardness of the zucchini,and the visible/near-infrared spectroscopy technique was used to detect the hardness of the zucchini.The results showed that the PLSR prediction model established by S-G+SNV processing was the best,the correlation coefficient of correction set was 0.979,and the correlation coefficient of prediction set was 0.976.The predictive correlation coefficient of validation model was 0.886,and the predicted root mean square error was 0.126.It was feasible that the visible/near infrared spectrum technique was used to predict the hardness index of zucchini.The study results will provide a theoretical basis for the rapid and non-destructive testing of fruit and vegetable hardness index.

关 键 词：可见/近红外光谱 硬度 西葫芦 无损检测 

分 类 号：S642.6[农业科学—蔬菜学]