论文部分内容阅读
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。
Aim To study the method to correct the content of volatile basic nitrogen (TVB-N) in pork freshness index, and to improve the applicability of spectral correction model to different kinds of pork samples. Methods A TVB-N model based on partial least squares regression (PLSR) was established to update the model by using spectral signal correction and model updating. The modified model was used to analyze the genetic diversity of Enshi mountain pigs Predict the effect. The results showed that the prediction coefficient R2p was 0.884 and the prediction standard deviation RMSEP was 1.792. The model was used to predict the TVB-N value of Enshi mountain pig with R2p of 0.552 and RMSEP of 4.733. The revised Du Chang model predicts Enshi pig TVB-N value, R2p increased to 0.964 and 0.943, RMSEP decreased to 1.329 and 1.885. Conclusion Both the spectral signal correction and the model updating method can effectively improve the model predictive performance and improve the model adaptability.