基于红外光谱角度度量模型同时测定辣椒油品质指标-《中国调味品》

文章信息/Info

Title:: Simultaneous Determination of Chili Oil Quality Index Based on Infrared Spectral Angle Measurement Model

摘要:: 本文利用向量的方向确定性，将光谱强度度量转换成空间角度度量，重构辣椒油红外光谱与酸值、折光率、过氧化值的线性关系，建立辣椒油红外光谱与关键品质指标的分析模型。共采集40个辣椒油样本红外光谱数据，用传统方法测定其三种指标真实值，对光谱进行不同预处理后，分别选取不同波段，利用自编角度度量算法(VAPLS)建立辣椒油三种指标预测模型，并与传统偏最小二乘法(PLS)进行对比。结果表明：酸值校正集和验证集相关系数为0.9936、0.9959，均方根误差0.0282、0.0264；折光率相关系数为0.9925、0.9907，均方根误差0.000132、0.000147；过氧化值相关系数0.9923、0.9421，均方根误差0.0176、0.0843。市售辣椒油三种指标均方根误差为0.0185、0.0001、0.0355。均优于PLS，说明角度度量所建模型有较好的预测性和稳定性。

Abstract:: In this paper, by using the direction determination of the vector, the intensity measurement of the spectrum is converted into spatial angle measurement, and the linear relationship between the infrared spectrum of chili oil and the acid value, refractive index, a peroxidation value of the chili oil is reconstructed, the analysis model of the infrared spectrum and key quality index of chili oil was established.The infrared spectral data of 40 samples of chili oil were collected, and the true values of the three indexes were determined by the traditional method.Selecting the different bands of pretreated spectrum, the self-programmed angle measurement algorithm (VAPLS) was used to establish the prediction model of three indicators of chili oil, and the model was compared with the traditional partial least squares (PLS) method.The results show that the correlation coefficient of the acid value correction set and the verification set were 0.9936,0.9959, and the root mean square error were 0.0282,0.0264, respectively.The refractive index correlation coefficient were 0.9925,0.9907, and the root-mean-square error were 0.000132,0.000147.The correlation coefficient of peroxidation value were 0.9923,0.9421, and the root mean square error was 0.0176.The results were all better than PLS, which indicated that the model of Angle measurement had better predictability and stability