梨枣产地的可见_近红外光谱判别研究.docVIP

  梨枣产地的可见/近红外光谱判别研究# 张淑娟，赵聪慧，杨国强，赵艳茹，尹丽华* （山西农业大学工学院，山西 太谷 030801） 5 10 15 20 25 30 35 40 摘要：不同产地梨枣的判别有利于消费者选购时溯源。利用可见/近红外光谱，对处于采收 期的来自三个不同产地的同一品种梨枣进行分类判别建模方法研究。采用不同的预处理方法 优化样本的光谱数据，基于全波段和短波近红外、中波近红外和长波近红外建模波段建立不 同产地梨枣的 LS-SVM 判别模型，比较各模型的识别率，确定最优的不同产地梨枣判别建 模方法。结果表明，对样本光谱数据一阶微分预处理之后，再用主成分分析法提取 20 个主 成分建立的短波近红外 LS-SVM 模型效果最优，正确判别率达到了 100% 关键词：可见/近红外光谱；最小二乘支持向量机；梨枣；产地；判别 中图分类号：S123；S665 Origin Discrimination of “Lizao” jujubes based on Visible / near-infrared spectroscopy ZHANG Shujuan, ZHAO Conghui, YANG Guoqiang, ZHAO Yanru, YIN Lihua (College of Engineering, Shanxi Agricultural University, ShanXi TaiGu 030801) Abstract: The identification of different origin “Lizao” jujube can help traceability in consumers purchase. Classification discrimination model of same varieties “Lizao” jujubes from three different origins were studied in the harvest period by using visible / near-infrared spectroscopy. The spectral data was optimized with different pretreatment methods, and different origin LS-SVM discriminantion model of “Lizao” jujubes were established in full-band near-infrared and shortwave, medium wave and long wave infrared band near-infrared, optimal discriminantion model of fresh jujubes from different areas was determined through compared recognition rate of each model. The results showed that short-wave near-infrared LS-SVM model, which spectral data was processed with first derivative, and 20 main components were extracted by principal component analysis is optimal with correct classification rate of 100%. Keywords: Visible / near-infrared spectroscopy; LS-SVM; ”Lizao” jujube; origin; discrimination 0 引言 根据现代药理研究，红枣有增强体能、加强肌力的功效，其中梨枣以果实大、皮薄肉厚、 清香甜脆、风味独特，越来越受到人们的重视和消费者的欢迎，是很有发展前途的优良鲜食 品种之一。由于气候、土壤、灌溉、管理等条件的不同，梨枣在不同产区其果实的营养成分 和品质存在一定的差异[1,2]。因此，对不同产地梨枣判别有利于消费者选购时溯源。 近红外光谱分析技术在农产品品质无损检测中得到了广泛应用，用近红外光谱对鲜枣品 质及品种等进行检测已有报道。张淑娟等[3,4]利用近红外漫反射光谱对鲜枣品种和可溶性固 形物含量进行检测；马毅等[5,6]研究不同预处理方法对 PLS 模型检测鲜长枣糖度的影响，将 小波变换去噪应用于鲜枣糖度近红外光谱检测的研究；J. Wang 等[7,8]利用近红外光谱结合 PLS 对鲜梨枣的可溶性固形物含量进行