摘要
以225个猪肉肉糜样本为研究对象,利用可见/近红外光谱分析技术对猪肉肉糜主要品质指标的的快速检测进行了研究。光谱经小波去噪后,采用偏最小二乘法和支持向量机定量分析方法分别建立了肉糜中肌内脂肪、蛋白质和水分含量的可见/近红外光谱预测模型。其中,肌内脂肪的支持向量机定量预测模型最优,校正相关系数rcal和预测相关系数rval为0.889和0.888;蛋白质的偏最小二乘定量预测模型最优,校正相关系数rcal和预测相关系数rval为0.869和0.881;水分的偏最小二乘定量预测模型最优,校正相关系数rcal为0.877,预测相关系数rval为0.848,所有模型的预测相对分析误差(RPD)均小于3.0。研究表明,可见/近红外光谱分析技术可用来检测猪肉肉糜品质,进一步提高所建模型的精度和稳定性可应用于实际检测。
The objective of the present study was to estimate minced pork meat quality using visible and near infrared （Vis-NIR） spectroscopy. Two hundred twenty five carcasses samples from longissimus dorsi muscle were scanned over the Vis-NIR spectral range from 350 to 1 015 nm and analysed for intramuscular fat （IMF）, protein and moisture according to the official methods. Wavelet transform was employed to eliminate the spectra noise. Partial least square regression （PLSR） and support vector machine （SVM） were used to develop Vis-NIR spectroscopy models for chemical composition detection. According to calibration statistics, the best model to predict intramuscular fat content was developed by SVM with the denoised spectra, the correlation coefficient was 0. 889 for calibration and 0. 888 for validation. For protein and moisture, the best model was achieved with the PLS method with the correlation coefficient of 0. 869 and 0. 881 for protein calibration and validation sets and 0. 877 and 0. 848 for moisture calibration and validation sets, respectively. And all the ratios of standard deviation of validation set to root mean square error of prediction （RPD） were not more than 3.0. Results indicated that it was possible to predict chemical composition in minced pork meat. As a fast predictor of meat quality using Vis-NIR spectroscopy, it is necessary to improve the precision and the robustness of the model for practice.
出处
《光谱学与光谱分析》
SCIE
EI
CAS
CSCD
北大核心
2011年第10期2734-2737,共4页
Spectroscopy and Spectral Analysis
基金
国家(863计划)高技术研究发展计划项目(2007AA10Z215)
浙江省自然科学基金项目(Y307441)资助
关键词
猪肉肉糜
可见/近红外光谱
小波变换
偏最小二乘回归
支持向量机
Minced pork
Visible/near infrared spectroscopy （Vis-NIR）
Wavelet transform
Partial least squares regression （PLSR）
Support vector machine （SVM）
