本文选择来自昌邑新昌、牟平仙坛、莱阳春雪、亚太中慧四个产地，不同厂家的180只同鸡龄同部位鸡肉样本，分别采用蒸、煮、微波三种方式加工后，经预处理粉碎过筛，利用安全、快速的近红外光谱技术，对不同产地的鸡肉样本进行近红外扫描并对所采集的近红外光谱进行聚类分析(CA)、主成分分析(PCA)，针对三种加工方式分别建立鸡肉产地溯源的定性判别模型，以探究加工后鸡肉产地溯源的可行性。试验表明：在波数范围7000~4000 cm-1内，原始光谱经二阶求导（13点平滑）和矢量归一化(SNV)预处理后，经三种方式加工后鸡肉的近红外光谱图均有显著差异，其中聚类分析判别正确率均高于90%，鸡肉样本的主成分空间分布位于不同的区域。利用来自四个厂区的独立样本经相同处理后对模型进行验证，识别正确率可达90%~95%。近红外光谱分析技术对加工鸡肉制品的产地溯源具有可行性。

One hundred and eighty chicken samples of the same age obtained from different factories within four production areas in Shandong were processed by three cooking methods (steaming, boiling, and cooking in a microwave), pretreated, crushed, and sieved. The samples were then scanned with a safe and rapid near-infrared (NIR) spectroscopic technique, followed by cluster analysis (CA) and principal component analysis (PCA) of the spectroscopic data, in order to establish a qualitative discrimination model for the traceability of cooked chicken. There were significant differences among the three cooking methods within the wavenumber range of 4000~7000 cm-1 after second derivative (13 point smoothing) and standard normal variable (SNV) pretreatment. The discrimination accuracy of CA was greater than 90%, while the principal components for the chicken samples were distributed in different geographical locations. The discrimination model, calibrated by using known samples, was used to predict unknown chicken samples, which attained a discrimination accuracy of 90%–95%. These results show that applying NIR spectroscopy to trace the geographical origins of cooked chicken is feasible.

山东省自主创新专项（2013CXC90202）