枇杷内部品质近红外光谱无损检测

Nondestructive detection of internal quality of loquat by near infrared spectroscopy

利用近红外光谱技术(near infrared spectroscopy,NIRS)对101个枇杷样品进行无损检测,测得样品的可溶性固形物(total soluble sdid,TSS)、可滴定酸和Vc含量,结合偏最小二乘法(partial least squares,PLS)分别建立TSS、可滴定酸和Vc含量的定标模型。采用定标模型分别对TSS、可滴定酸和Vc的验证集样品进行预测,预测决定系数Rp2分别为0.906,0.745,0.554,预测均方根误差(root-meansquare error of prediction,RMSEP)分别为0.628,0.048,2.230,且TSS的相对分析误差(relative prediction deviation,RPD)为3.31,可滴定酸和Vc的RPD分别为2.00,1.52。表明建立的枇杷TSS的定标模型可用于实际检测,枇杷的可滴定酸和Vc含量可采用近红外光谱进行检测,但检测精度有待于进一步的提高。

Spectra data of the 101 loquat samples were obtained by using NIRS and the TSS,titratable acid,Vc of the samples were determined by conventional chemical methods.The mathematical pretreatment methods were used as spectral preprocessing options,and then the calibration models of TSS,titratable acid and Vc were established respectively by PLS.By using these models,the TSS,titratable acid and Vc in the validation sets can be predicted.The Rp2 are 0.906,0.745,0.554,RMSEP are 0.628,0.048,2.23 respectively,and the RPD of TSS is 3.31.Thus,the calibration model of TSS can be used for the actual detection.The RPD of titratable acid and Vc are 2and 1.52 respectively.The content of titratable acid and Vc in the loquat can be tested by NIRS.However,the detection accuracy remains to be improved.