近红外光谱在小尾寒羊羊肉水分预测中的应用

Application of Near Infrared Spectroscopy in Water Prediction of Small-Tail Han Sheep

以南疆小尾寒羊鲜羊肉的含水量为研究对象,首先采集鲜羊肉的光谱数据和水分含量信息;然后分别采用小波变换、多元散射校正以及二者结合的方法预处理数据;最后使用偏最小二乘法对3种方法预处理过的光谱数据建立羊肉水分含量的预测模型,共制备134个样本,根据留一法选择110个样本作为训练集,剩余的24个样本作为预测集。结果表明,采用多元散射校正预处理方法建立的模型预测能力优于小波变换,采用2种结合的预处理方法建立的模型最优,预测精度达到0.98972,相关系数达到0.93807,预测均方差为0.0094793。这说明基于小波变换与多元散射校正的数据预处理方法结合近红外光谱技术,对南疆小尾寒羊鲜羊肉的水分含量使用偏最小二乘法建立预测模型是可行的,可为研制鲜羊肉水分含量实时监测设备提供理论依据和指导。

In this paper,the water content of fresh mutton from Small-Tailed Han Sheep in Southern Xinjiang was studied.Firstly,spectral data and moisture content information of fresh mutton were collected.Then wavelet transform,multiple scattering correction and the combination of the two methods were used to preprocess the data.At last,partial least square method was used to establish the prediction model of mutton moisture content with the spectral data pretreated by the three methods.A total of 134 samples were prepared.According to the retention method,110 were selected as the training set and the remaining 24 as the prediction set.The results showed that the prediction ability of the model was better than that of the wavelet transform,and the best model was established by combining the two pretreatment methods.The prediction accuracy was 0.98972,the correlation coefficient was 0.93807,and the mean square error was 0.0094793.This showed that it was feasible to preprocess the spectral data with wavelet transform and multiple scattering correction,and to use partial least square method to establish a prediction model for the spectral data and the moisture content of Small-Tailed Han Sheep.It can provide theoretical basis and guidance for the development of water content monitoring equipment for fresh mutton.