猪肉中挥发性盐基氮含量光谱检测模型的修正方法

Correction methods of pork total volatile basic nitrogen content detectionmodel based on hyperspectral imaging technology

中文关键词:  模型修正  猪肉  挥发性盐基氮  高光谱  偏最小二乘回归

英文关键词:model correction methods  pork  total volatile basic nitrogen  hyperspectral imaging technology  partial least squares regression

中文摘要:

      目的 研究猪肉新鲜度指标挥发性盐基氮(TVB-N)含量检测模型修正方法, 以提高光谱校正模型对不同品种猪肉样品的适用性。方法 建立基于偏最小二乘回归(PLSR)的杜长大猪肉TVB-N模型, 采用光谱信号补正与模型更新两种方法对该模型进行修订, 比较修正后杜长大模型对恩施山猪样本的预测效果。结果 建立的杜长大猪肉样本模型预测决定系数R2p为0.884, 预测标准差RMSEP为1.792, 将此模型用于预测恩施山猪TVB-N值, R2p为0.552, RMSEP为4.733。修正后的杜长大猪肉样本模型预测恩施山猪TVB-N值时, R2p分别提高到0.964和0.943, RMSEP分别降低为1.329和1.885。结论 光谱信号补正和模型更新方法均能有效改善模型预测性能, 提高模型适应性。

英文摘要:

      Objective To study correction methods for pork freshness (TVB-N) detection model of different species based on hyperspectral imaging technology and improve the generality of the calibration model. Methods Du changda model was established based on partial least squares regression using Du changda mountain boars as samples. Model updating by adding new typical samples and spectral correction based on model regression coefficient were adopted to improve the model applicability of the calibration model for Enshi mountain boars. Results The TVB-N content model, with 0.884 as the coefficient of determination in prediction sets (R2p) and 1.792 as the root mean squared error of prediction (RMSEP), was used to predict the Enshi mountain boars, and R2p and RMSEP were 0.552 and 4.733, respectively. While the R2p increased to 0.964 and 0.943 and the RMSEP decreased to 1.329 and 1.885 using calibration model. Conclusion Both methods can improve the predict performance of model effectively, and enhance the model adaptation.