一种便携式苹果糖度无损检测仪的研制

摘要

糖度是判断苹果质量好坏的一个重要参考标准,针对苹果糖度的检测问题,设计了一种以Cortex-A9为内核、以自研发的可见-近红外光谱仪(波长范围400~1 000 nm)作为光谱检测装置、以Linux为操作系统的便携式苹果糖度无损检测仪.以山东烟台的100个红富士苹果为材料,采集了漫透射检测方式下基于自收发光机构的苹果漫透射光谱曲线,结合化学计量学方法,对样本的全光谱曲线使用了平均法和Savitzky-Golay卷积平滑光谱预处理方法,将预处理后的光谱数据按波峰位置划分区间,并分别按照全光谱范围和所划分区间的波段范围建立PLS模型来预测苹果的糖度含量.结果表明,经预处理后的全光谱数据所建立的PLS模型预测效果最好,优于按波峰划分区间所建立的PLS模型,其校正相关系数为0.96、预测相关系数为0.87,校正均方根误差为0.31、预测均方根误差为0.34.同时对仪器工作时的预测稳定性进行了测试,测试结果得出检测精度可控制在±0.2 Brix以内,模型预测精度满足现场快速检测应用要求.%Sugar content is an important reference standard to judge the quality of apple,a portable apple soluble solids content spectrometer was designed with Cortex-A9 as controller,self-developed near-infrared spectrometer(the wavelength range of 400 to 1 000 nm) as spectrum detector,Linux as embedded operating system.The total number of 100 apple samples from Yantai,Shandong Province were used as the test objects.Collecting optical transceiver diffuse transmission spectral curves,combined with the method of chemometrics,the full spectrum of the sample was processed by averaging method and Savitzky-Golay convolution smoothing spectral pretreatment method,and the spectral data of pretreatment were divided according to the wave peak position,then PLS model was established to predict the sugar content of apple.Experimental results showed that the full spectrum of PLS models set up in average and smooth pretreatment way were better than that of divided by wave peak position.The correlation coefficient of Rc=0.96,the Rp=0.87,root mean square error RMSEC=0.31,RMSEP=0.34.Meanwhile,work to predict the stability of the instrument were tested,the test results obtained detection accuracy could be controlled within ± 0.2Brix,the prediction accuracy of the model met the requirements of on-site inspection applications.