基于MATLAB软件的苹果内部品质系统化无损检测

Systematic non-destructive test of the apple internal quality based on MATLAB software

研究将近红外设备与MATLAB软件链接,实现苹果糖度及可滴定酸度的近红外光谱扫描、数据处理、模型建立、含量预测及模型评价等在线无损检测水果内部品质的一体化。以市售陕西红富士苹果为材料,通过近红外文件格式转换、近红外光谱预处理、偏最小二乘法(PLS)回归分析、预测模型的建立、结果预测及模型精度检验等工作编写和调试基于MATLAB语言的模块化程序,经主程序调用建立苹果糖度及可滴定酸度的定量预测模型。结果表明:本实验所建预测模型糖度及可滴定酸度预测值和真实值之间的相关系数R分别为0.9528,0.8786;标准校正误差(SEC)分别为0.4788,0.0215;标准预测误差(SEP)分别为0.3170,0.0128;标准偏差(SD)分别为1.4111,0.0390,模型具有较高的预测精度。因此,将近红外设备与MATLAB软件链接,实现苹果糖度及可滴定酸度在线无损检测的一体化具有可行性且模型检测精度较高,对开发水果内部品质评价及在线分级软件具有重要意义。

The objective of this study was through linking nearly infrared equipment with MATLAB software to realizeonline non-destructive testing the fruit internal quality based on integration of infrared light scanning, dataprocessing, prediction model and model evaluation of apple sugar and titratable acidity. Through writing anddebugging the MATLAB modularized program of the near infrared spectra pretreatment, partial least squares（PLS）regression analysis, quantitative prediction model, prediction results and model accuracy test to get thequantitative forecasting and model evaluation of apple sugar and titratable acidity .The results showed that thecorrelation coefficient（R）between predicted values and real values,Standard error of correction（SEC） ,Standarderror of prediction（SEP） ,and the standard deviation（SD） of apple sugar（titratable acidity）builded the predictionmodels were 0.9528 （ 0.8786 ）, 0.4788 （ 0.0215 ）, 0.3170 （ 0.0128 ）, and 1.4111 （ 0.0390 ）, respectively .it had beendemonstrated that rapid quantitative forecasting the content of apple sugar and titratable acidity using NearInfrared reflectance spectroscopy based on MATLAB 7.1 were feasibility, and the model had a good predictionaccuracy.The method of quantitative online non-destructive prediction the content of apple sugar and titratableacidity will have a important significance to develop internal quality automation evaluation software of fruit andclassification software.