基于小波变换的方便面含油率近红外光谱检测技术

Improvement on Predicting Precision of Oil Content in Instant Noodles with Near-infrared Spectroscopy Treated by Wavelet Transformation

以检测方便面的含油率为例 ,探讨了用小波变换原理分析近红外漫反射光谱并提取有效信息的方法。结果表明 ,变换后的光谱更能反映出与含油率之间的相关关系。通过与使用方便面原始漫反射光谱、一阶导数、二阶导数的多元回归分析方法比较 ,得到使用 8个尺度变换后的小波系数与含油率之间的关系最为显著 ,且四元回归分析的预测精度最好 ,比采用二阶导数时的平均相对预测误差降低 1.1个百分点 ,相对误差的标准差降低0 .18。

Near infrared spectroscopy(NIRS),with the characters of quickness, non destruction, long detection range, high precision and reliable detection data, is a new and popular method for quantitative and qualitative analysis. Whether NIRS can be used in practical production or not, it depends on its predicting precision. The oil content of instant noodles is inspected by NIRS analyzed using wavelet transformation principle. The results show that the spectra treated with wavelet transformation indicate the relationship with oil content in instant noodles more effectively. Compared with initial, first and second derivatives spectra, wavelet transformation of 8 size has the most remarkable relation with oil content. The predicting precision of 4 element regression is the best, whose relative error lowers from 5 901% and 5 741% to 4 657%, the standard error lowers from 1 921 and 1 842 to 1 665 than those of initial and second derivative spectra respectively.