研究了不同采集状态的虾样品对近红外光谱PLS模型的影响。利用DA7200近红外光谱仪,采集南美白对虾完整虾和虾糜的近红外光谱曲线。采用Unscrambler10.3软件选择最佳光谱预处理方法和最优波段,建立了完整虾和虾糜与挥发性盐基氮(TVB-N)值、菌落总数(TBC)值关联的偏最小二乘(PLS)模型,并对模型进行评价和验证。结果表明:定标集虾糜模型中的校正相关系数rc,校正决定系数Rc2,交叉验证相关系数rv,交叉验证决定系数Rv2,均高于完整虾模型;校正均方根误差RMSEC,校正标准误差SEC,交叉验证均方根误差RMSECV,交叉验证标准误差SECV均低于完整虾模型。验证模型中虾糜预测模型中相关系数r均大于完整虾预测模型,预测均方根误差RMSEP,预测标准误差SEP均低于完整虾预测模型,且虾糜预测模型对TVB-N、TBC值预测值更为准确,表明以虾糜作为近红外光谱采集状态优于完整虾。

The influence of near infrared spectral PLS model on different shrimps was analyzed.The near infrared spectral(NIS) curve of whole shrimp and minced shrimp was collected by near infrared spectroscopy analyzer(DA7200).The Unscrambler 10.3 software was used to choose the best preprocessing method and the optimal band to establish TVB-N and TBC models by partial least square regression(PLSR),and validate model.The results showed that the correlation coefficient(rc),the correction of decision(Rc2),the correlation coefficients of cross-validation(rv),the correction of decision of Cross-Validation(Rc2) of minced shrimp models were higher than those of whole shrimp models in calibration set.The root mean square error correction(RMSEC),standard error of calibration(SEC),root mean standard error of cross-validation(RMSECV),standard error of cross-validation(SECV) of minced shrimp models were not higher than those of whole shrimp models in calibration set.The correlation coefficients,the root mean square errors of prediction(RMSEP) and standard error prediction(SEP) of different models showed that the results of minced shrimp models were better than whole shrimp models in prediction set.And the minced shrimp models were better predict TVB-N and TBC in Penaeus vannamei Boone than whole shrimp model.In conclusion,minced shrimp NIR curve was better in prediction than the whole shrimps.