论文部分内容阅读
目的:应用近红外光谱技术快速检测鹅肉的嫩度值。方法:采集完整鹅肉的近红外光谱(950~1 650 nm),光谱经多种校正预处理后,再分别采用主成分回归和偏最小二乘法建立鹅肉嫩度的定量预测数学模型。结果:采用5点移动窗口平滑处理结合偏最小二乘法所建立模型的预测效果最好,嫩度定量校正数学模型的模型决定系数为0.908 0,内部交互验证均方根误差为113.618 6。用此模型对预测集20个样品进行预测,预测值与实测值的相关系数达到0.971 1,预测值平均偏差为21.673 g,预测值和实测值之间没有显著性差异(P>0.05)。结论:近红外光谱作为一种无损快速的检测方法,可用于评价鹅肉的嫩度。
Objective: To quickly detect the tenderness of goose by using near infrared spectroscopy. Methods: The complete goose meat was collected by near infrared spectroscopy (950-1 650 nm). After the spectra were preprocessed by various calibrations, quantitative prediction mathematical model of goose tenderness was established by principal component regression and partial least squares respectively. Results: The model with 5-point moving window smoothing combined with partial least squares method had the best prediction performance. The model determination coefficient of the mathematic model for quantitative correction was 0.908 0, and the root mean square error of internal interaction verification was 113.618 6. The prediction of 20 samples from the prediction set was carried out by this model. The correlation coefficient between the predicted value and the measured value reached 0.971 1, and the predicted average deviation was 21.673 g. There was no significant difference between the predicted value and the measured value (P> 0.05). Conclusion: Near infrared spectroscopy as a nondestructive and rapid detection method can be used to evaluate the tenderness of goose.