为了实现香梨多种营养成分指标高效且较为准确的检测，本文采用质构仪对香梨果肉进行了蠕变测试和应力松弛测试，确定最适合描述香梨果肉蠕变特性和应力松弛特性的流变模型以获取香梨各流变参数，然后分析流变参数与香梨可溶性固形物含量SSC、可滴定酸度TA、固酸比RTT的相关性，并采用逐步多元线性回归法预测香梨这3种营养成分指标。研究结果表明，香梨果肉的蠕变特性和应力松弛特性分别可采用六元件广义Kelvin-Voigt模型和五元件广义Maxwell模型精确拟合，决定系数R2均在0.95以上；两模型都有流变参数与香梨可溶性固形物含量SSC、可滴定酸度TA、固酸比RTT具有显著相关性（p<0.05），所构建的逐步多元线性回归模型也都可以有效预测这3种营养成分指标，预测模型的相关系数r均高于0.7，所得预测值与实测值间无显著差异。与应力松弛参数构建的预测模型相比，基于香梨蠕变参数构建的预测其各营养成分指标模型的标准误差RMSE较小，对香梨营养成分指标预测更稳定可靠。

To achieve efficient and accurate detection of various nutritional composition indexes of Korla pear flesh, creep-recovery and stress relaxation tests were performed on pear flesh with using a TA-XT plus Texture Analyzer. Rheological models which were the fittest for describing Korla pear flesh’s creep-recovery and stress relaxation properties were determined, and the rheological parameters of Korla pear flesh were obtained with the models. Then, the correlations between the rheological parameters and soluble solid content (SSC), titratable acidity (TA), solid acid ratio (RTT) of pear flesh were analyzed, and the three nutritional composition indexes were predicted through stepwise multi-linear regression analyses. The results showed that creep-recovery and stress relaxation properties of Korla pear flesh can be best characterized by a generalized Kelvin-Voigt model with six elements and a generalized Maxwell model with five elements respectively, and the coefficients of determination R2 were higher than 0.95. Partial rheological parameters of the two models were significantly correlated with the SSC, TA and RTT (p<0.05), hence, the constructed stepwise multi-linear regression models could also effectively predict the three nutritional composition indexes. The correlation coefficient r of the prediction models were higher than 0.7, and there was no significant difference between the measured values and the predicted values. The RMSE values of the models for predicating the three nutritional compositions based on creep parameters were smaller than those of the models based on stress-relaxation parameters. Therefore, the models constructed on the basis of creep parameters were more stable and reliable for predicting nutritional compositions of Korla pear flesh.

国家自然科学基金地区科学基金项目（31560476）