近红外光谱鉴别滇南小耳猪和DLY商品猪

摘要

Intact, ground, oven-dried and freeze-dried pork samples from Diannan small-eared pigs and DLY commercial pigs were analyzed for breed discrimination using spectral modeling by near infrared relfection spectroscopy (NIRS). In order to select the optimal spectral modeling method, the spectra of intact, ground, oven-dried and freeze-dried samples of longissimus dorsi from 53 Diannan small-eared pigs and 58 DLY commercial pigs in the range of 4 300–11 000 cm-1 were acquired, and the original spectra were preprocessed by ifrst derivative, second derivative, standardization of variables, and multiple scattering correction, respectively. The models for breed discrimination were established using soft independent modeling of class analogy (SIMCA). Eight of the 111 samples were selected as calibration set, while the remaining 31 were included in validation set. Results showed that the recognition rate and rejection rate for DLY commercial pig in the validation set by the model of intact pork were 100% and 0, and those for Diannan small-eared pig in the validation set were 100% and50%, respectively, which were unsatisfactory. Similarly, unsatisfactory recognition rate and rejection rate for both pig breeds in the validation set were obtained using the model of ground pork, which were 100% and 0, respectively. However, the models of oven-dried and freeze-dried pork exhibited good recognitionperformance, using which the recognition rate and rejection rate for the validation set were both 100%. To sum up, NIRS is useful to discriminate between Diannan small-eared pigs and DLY commercial pigs.%应用近红外漫反射光谱技术（near infrared reflectance spectroscopy，NIRS），对滇南小耳猪和DLY商品猪的整块、匀质肉糜、烘干粉、冻干粉的光谱建模分析，进行品种鉴别，并筛选出最优的光谱建模方法。采集4300～11000 cm-1范围内53头滇南小耳猪、58头DLY商品猪背最长肌整块、匀质肉糜、烘干粉、冻干粉光谱数据。分别采用原始光谱、一阶导数、二阶导数、变量标准化、多元散射校正等方法进行光谱预处理，应用簇类独立软模式方法建立品种鉴别模型。其中80个样品作为校正集，余下31个样品作为验证集。结果表明：DLY商品猪和滇南小耳猪整块肉光谱模型对验证集DLY商品猪识别率为100%，拒绝率为0%，对验证集滇南小耳猪识别率为100%，拒绝率50%，效果不理想；匀质肉糜光谱模型对验证集DLY商品猪识别率为100%，拒绝率为0%，对验证集滇南小耳猪识别率为100%，拒绝率0%，效果不理想；烘干粉、冻干粉的近红外光谱模型鉴别效果较好，验证集的识别率和拒绝率均为100%。综上，近红外光谱分析方法可用于鉴别DLY商品猪和滇南小耳猪两个品种。