基于近红外光谱的芝麻油酸价含量的预测

采用近红外光谱分析技术对芝麻油的酸价含量进行检测,避免了传统的化学方法缺陷,同时在不破坏样品的前提下极大地提高了检测效率。对39个芝麻油样本的酸价光谱图进行光谱预处理优化,并选择适当的光谱范围,采用偏最小二乘法(PLS)和BP神经网络算法进行了定量分析研究。结果表明,在所选定的样本和光谱范围内,PLS和BP神经网络算法均可以用于芝麻油酸价含量的预测,采用PLS模型的预测均方根误差(RMSEP)为0.058;用BP神经网络预测的RMSEP为0.148 8,偏最小二乘法建模相对于一般的BP网络建模方法更具有较好的建模预测效果。

Detection of acid value in sesame oil based on near infrared spectrum

The technology of near infrared spectral analysis was used to test the acid value in sesame oil. The drawbacks of traditional chemical meth- ods were avoided and the detection efficiency was greatly improved under the premise of undamaging samples. The acid value spectra of 39 sesame oil samples were optimized for pretreatment and the appropriate spectrum was selected, in addition, the partial least squares （PLS） and BP neural network algorithm for quantitative analysis of the research was conducted. Result showed that in the selected samples and spectrum range, the PLS and BP neural network algorithm could be used for sesame oil acid value prediction. The root mean square error of prediction （RMSEP） of PLS was 0.058, while the RMSEP of BP was only 0.148 8. Result proved that the PLS had better modeling prediction effect than the BP neural network algorithm.