论文部分内容阅读
通过采集市售生鲜鸡肉样品在中波近红外区的光谱信息,采用偏最小二乘法构建胆固醇定量分析模型,并评价模型的预测准确性。在建模过程中,讨论异常样品剔除与组合预处理方法等优化措施对模型的影响。参模样品经2次异常样本剔除并以SG(Savitzky-Golay)一阶导数、SG平滑及去趋势校正进行预处理后,获得了最佳生鲜鸡肉胆固醇定量分析模型,其中:校正标准差(standard error of calibration,SEC)为4.534 8、校正集相关系数为0.919 7、预测标准差(standard error of prediction,SEP)为7.437 5、验证集相关系数为0.812 0、范围误差比为2.844 7、相对预测标准差为9.44%、主因子数为5、SEP/SEC为1.640 1。对检验集样品预测的结果表明,基于中波近红外光谱构建的胆固醇定量分析模型的预测性能较好(P>0.05),特别是在60~100 mg/100 g含量区间,可应用于对市售生鲜鸡肉及产品胆固醇的检测。
By collecting spectral information of commercial fresh chicken samples in the mid-infrared and near-infrared region, the partial least squares method was used to construct the quantitative analysis model of cholesterol, and the prediction accuracy of the model was evaluated. In the process of modeling, the influence of optimization measures such as abnormal sample rejection and combination pretreatment methods on the model is discussed. The samples of ginseng samples were removed after 2 abnormal samples and were pretreated with the first derivative of SG (Savitzky-Golay), SG smoothing and defoliation correction, and the best quantitative model of cholesterol in fresh chicken was obtained. The calibration standard deviation the standard error of calibration (SEC) was 4.534 8, the correlation coefficient of calibration set was 0.919 7, the standard error of prediction (SEP) was 7.437 5, the correlation coefficient of validation set was 0.812 0, and the range error ratio was 2.844 7. The standard deviation of prediction was 9.44%, the number of major factors was 5 and SEP / SEC was 1.6401. The predicted results of the test samples showed that the predictive performance of the quantitative cholesterol analysis model based on the microwave near-infrared spectroscopy was better (P> 0.05), especially in the range of 60-100 mg / 100 g, Sales of fresh chicken and cholesterol testing.