基于近红外光谱技术的鲢鱼营养成分的快速分析

Fast Analysis on Nutrient of Silver Carp Based on Near Infrared Spectroscopy

中文关键词:  近红外光谱  鱼  营养成分  偏最小二乘  主成分分析  人工神经网络

英文关键词:Near Infrared Spectroscopy (NIRS)  Fish  Nutrient  Partial least square (PLS)  Principal component analysis (PCA)  Artificial neural network (ANN)

中文摘要:

      目的 通过采集鲢鱼的近红外光谱数据和测定鱼肉营养成分含量探索鲢鱼营养成分的快速分析方法。方法 采集254个鲢鱼鱼肉样品的近红外光谱数据, 经过多元散射校正、正交信号校正、数据标准化等20种方法预处理, 在1000~1799 nm光谱范围内, 结合化学实测值分别采用偏最小二乘法、主成分分析和BP人工神经网络技术、偏最小二乘法和BP人工神经网络技术建立鲢鱼营养成分近红外定量模型。结果 鲢鱼鱼肉粗蛋白含量为12.05%~19.05%, 粗脂肪含量为0.24%~5.27%, 水分含量为72.62%~80.58%, 灰分含量为0.46%~1.50%, 数据范围较大, 可满足建模要求。在3种建模方法中, 近红外光谱数据结合偏最小二乘法建立的鲢鱼营养成分模型最优, 所得的粗蛋白、粗脂肪、水分和灰分的近红外定量模型的相关系数分别为0.9969、0.9925、0.9831和0.9976。结论 采用近红外光谱数据和偏最小二乘法建立的模型具有较好的预测能力, 能较为准确、快速地分析出鲢鱼鱼肉粗蛋白、粗脂肪、水分和灰分的含量。

英文摘要:

      Objective To explore a rapid analysis method for nutrient of silver carp through collecting near infrared spectroscopy and determining nutrient. Methods The near infrared (NIR) spectra of 254 silver carp samples were collected. The diffuse reflectance spectra of samples were performed with different spectral pretreatments, such as multiplicative scatter correction (MSC), orthogonal signal correction (OSC), and standardization (S). The near infrared quantitative analysis models were obtained by partial least square (PLS) regression, principal component analysis (PCA) combined with back propagation artificial neural network (BP-ANN), partial least square combined with back propagation artificial neural network with 1000~1799 nm, respectively. Results The results showed that the protein content of silver carp ranged from 12.05% to 19.05%, the fat content from 0.24% to 5.27%, the moisture content from 72.62% to 80.58%, and the ash content from 0.46% to 1.50%. The nutrient measured values met the modeling requirements. The analysis models obtained by PLS were the best. The correlation coefficients of the models were 0.9969, 0.9925, 0.9831 and 0.9976 for protein, fat, moisture and ash content, respectively. Conclusion The results indicated that the models exhibited an acceptable fitting accuracy and predictive ability of analysis of protein, fat, moisture and ash content of silver carp by NIRS.