目的建立近红外光谱法结合偏最小二乘法测定许氏平鲉鱼肉中的脂肪和水分含量,以期简便、快速地对许氏平鲉进行品质分析与评价。方法采用常规分析手段测定70个样品的脂肪和水分含量,同时采集其近红外光谱数据,结合偏最小二乘法(partial least square,PLS)建立许氏平鲉鱼肉中脂肪和水分的定量预测模型,并对比不同光谱预处理方法、光谱范围和因子数对定量预测模型的影响。结果光谱经Savitzky-Golay(S-G)和标准正态变量变换(standardized normal variate,SNV)预处理后,在5341.85~4007.36 cm~(-1)、6556.79~5345.71cm~(-1)和8651.10~7162.33 cm~(-1)光谱范围内,选取主因子数10,建立脂肪的校正模型性能最优;光谱经过SNV预处理后,在8886.38~4061.35cm~(-1)光谱范围内,分别选取主因子数为9时,建立的水分的校正模型性能最优。脂肪和水分含量相对最优PLS模型的校正集相关系数分别为0.9918和0.9912,校正标准偏差分别为0.2680和0.3300,交叉验证相关系数分别为0.9820和0.9810,交叉验证均方差分别为0.3980和0.4850,验证集相关系数分别为0.9804和0.9798,验证集均方差分别为0.3260和0.3070。结论本方法可较为准确地预测许氏平鲉鱼肉中的脂肪和水分含量,能够满足快速分析评价许氏平鲉品质的要求。 OBJECTIVE To establish a method for the determination of fat and water content in Houttuynia harris f. Meat by near infrared spectroscopy combined with partial least squares in order to analyze the quality of Houttuynia harpinii easily and quickly. Methods The fat and water of 70 samples were determined by conventional analytical methods. The data of NIRS were collected and the PLS (partial least squares) method was used to establish the quantitative prediction model of fat and moisture of flesh. The effects of different spectral pretreatment methods, spectral ranges and factors on the quantitative prediction model were compared. Results After the spectra were pretreated by Savitzky-Golay (SG) and standardized normal variate (SNV), the spectra were in the range of 5341.85 ~ 4007.36 cm -1, 6556.79 ~ 5345.71 cm -1 and 8651.10 ~ 7162.33 In the spectrum of cm ~ (-1), the main factor of 10 was selected to establish the optimal calibration model of fat. After SNV pretreatment, the main factors were selected in the spectral range of 8886.38 ~ 4061.35cm ~ (-1) When the number is 9, the established water calibration model has the best performance. Correlation coefficients of relative fat and water content in the optimal PLS model were 0.9918 and 0.9912, respectively. The standard deviation of calibration was 0.2680 and 0.3300 respectively. The correlation coefficient of cross validation was 0.9820 and 0.9810 respectively. The mean square error of cross validation was 0.3980 and 0.4850 respectively. Set of correlation coefficients were 0.9804 and 0.9798 respectively, and the mean square error of validation set was 0.3260 and 0.3070 respectively. Conclusion The method can predict the fat and water content of Hübner’s flatfish fish more accurately and meet the requirements of rapid analysis and evaluation of Hübner’s flatfish quality.