中文摘要：

在前期探明苹果酒特征香气种类的基础上，对苹果酒发酵过程中特征香气的近红外光谱检测模型进行了研究，结果表明选择波数为9747．1—7498．3cm^-1和6102．0—5446．3cm^-1两段谱区，采用一阶导数和多元散射校正处理光谱后，利用偏最小二乘法建模效果较好，其中校正集R^2为O．9205，交叉验证均方根差为4．87mg／L；验证集预测值与实测值的R^2为0．9388，预测均方根差为3．76mg／L，表明利用近红外光谱法建立的模型达到了良好的预测效果。研究了苹果酒发酵过程中特征香气的产生特性，基于Luedeking—Piret方程，建立了描述苹果酒特征香气生成的动力学模型（R^2为0．9930），经检验表明该模型能够很好地拟合苹果酒发酵过程特征香气的生成状况。

英文摘要：

In order to measure and monitor character aroma content accurately and rapidly, calibration models were established during cider fermentation based on near infrared spectroscopy. The results indicated that the calibration models developed with partial least square （PLS） by the spectral pretreatment of first derivative （FD） and multiplicative scatter correction （MSC） in the characteristic absorption spectra ranges of 9 747. 1 - 7 498.3 cm-1 and 6 102.0 - 5 446.3 era-1 were the best. The correlation coefficients （R2） and the root mean square errors of cross validation of the model in calibration set were 0. 920 5 and 4. 87 mg/L, respectively. The R2 of the test set and the root mean square error of prediction were 0. 938 8 and 3.76 rag/L, respectively. These demonstrated that the model was very well and can be applied to quick determination and monitoring of character aroma content during cider-making. Based on the NIRS model, kinetic model of character aroma production was constructed with a R2 of 0. 993 0 by Luedeking - Piret equation. The results showed that the fitted values by the model were in an agreement with the observed values.