基于多光谱成像技术的牛肉干水分含量快速无损检测研究

Research on Rapid Nondestructive Detection of Moisture Content of Beef Jerky Based on Multispectral Imaging Technology

基于多光谱成像技术对牛肉干中水分含量的快速无损检测方法进行研究,通过对比最小二乘回归(PLS)、最小二乘支持向量机(LS-SVM)和误差反向传播神经网络(BPNN)所建预测模型的性能,发现BPNN模型对牛肉干水分含量预测效果最佳,其确定系数(R p 2)、预测集均方根误差(RMSEP)和剩余预测偏差(RPD)分别为0.941、3.602%和4.142。结果表明,光谱吸收度是检测牛肉干水分含量的重要特征,BPNN结合多光谱建立的预测模型精度较高,鲁棒性较好,在牛肉干水分的实时无损检测中具有良好的应用前景。

The rapid and nondestructive detection of moisture content in beef jerky based on multispectral imaging was performed.By comparing the results of different chemometrics methods such as partial least square(PLS),least square-support vector machine(LS-SVM)and back propagation neural network(BPNN),the best model was from BPNN method with the determination coefficients(R p 2),the root mean square error of prediction(RMSEP)and the residual prediction deviation(RPD)was 0.941,3.602%and 4.142.The results showed that spectral absorbance was an important feature for detecting the moisture content of beef jerky.The prediction model established by BPNN combined with multispectral had high accuracy and good robustness.It had a good application prospect in the real-time nondestructive detection of beef jerky moisture.