中文摘要：

利用近红外透反射光谱技术，研究短波近红外光谱（780～1100nm）无损检测蔗糖溶液的可行性，并通过主成分回归（PCR）和偏最小二乘（PLS ）方法建立了蔗糖溶液的近红外定量分析模型。采用Savitzky-Golay卷积平滑（5点）和多元散射校正（MSC）进行预处理，并且对预处理后的数据进行建模分析。PCR定量分析的结果：主成分数 PC ＝7，交互验证相关系数RCV ＝0.957335，交互验证的校正标准偏差RMSECV＝0.015859；PLS定量分析结果：主成分数PC＝4，交互验证相关系数RCV ＝0.975789，交互验证的校正标准偏差RMSECV＝0.012251。分别用PCR和PLS的校正模型对预测集样本进行预测，两种模型的预测标准偏差RMSEP分别为0.0127，0.0118。二者均对高浓度蔗糖溶液的预测结果比较理想，而且在PLS模型下，77％以上的样本相对误差在10％以下，较PCR模型的高。综合结果，PLS所建立的模型简单，而且精度很高，所以，基于短波近红外光谱的蔗糖浓度的快速无损检测是可行的。

英文摘要：

Near-infrared spectra of trans-reflective is applied to determine the feasibility of nondestructive testing of the sucrose solution in the near-infrared wavelength range of 780~1 1 00nm,and principal component regression （PCR） and partial least squares（PLS）are used to establish the near-infrared quantitative analysis model of sucrose solu-tion.After the pretreatment of （Savitzky-Golay）5 points smoothing and multiplicative scatter correction （MSC）,the （PCR and PLS）calibration model is made.The results of quantitative analysis of PCR are the number of principal component PC=7,the correlation coefficient RCV =0.957335,root mean square error of cross validation RMSECV=0.01 5859;The results of quantitative analysis of PLS are the number of principal component PC=4,the correlation coefficient RCV =0.975789,root mean square error of cross-validation RMSECV=0.01 2251.The prediction samples are predicted respectively by the calibration model of PCR and PLS.Root mean square errors of prediction are 0.01 27,0.01 1 8 respectively.Both of the methods are satisfactory for the predicted results of high concentration sam-ples.The relative errors of more than 77% of validation samples are below 1 0% using PLS.All of results show that PLS model is simpler and its prediction accuracy is higher,and short-wave near-infrared spectrometry is a valuable, rapid and nondestructive tool for the quantitative analysis of sucrose solution.