【摘要】 针对不同花椒品种的快速鉴别方法进行研究,以花椒的气味信息检测为研究对象,利用自行研制的电子鼻系统采集了6类花椒样品气味数据,对这些数据样本进行特征提取,得到了56组训练样本和32组测试样本。利用BP神经网络、概率神经网络和支持向量机对特征数据进行鉴别,正确识别率分别为89.58%、93.23%、94.27%,相对于BP神经网络和概率神经网络识别,支持向量机具有更好的分类效果。本文研制的电子鼻系统能无损、快速、准确鉴别花椒的品种,为农产品无损检测的研究提供了一种新的思路。更多还原

【Abstract】 A fast discriminating mehtod of Zanthoxylurn Bungeanum varieties was studied. In the experiment,the order information of six Zanthoxylurn Bungeanum varieties were collected by the electrocinc nose system for investigation. The fifty-six groups training samples and thirty-two groups testing samples were obtained after feature extraction. The feature data were classified by the BP neural network,probabilistic neural network,and support vector machine respectively,and the rate of discrimination was 89. 58%,93. 23%,94. 27%. Compared to the BP neural network and probabilistic neural networks,support vector machine had better classification results. The electronic nose system developed in this paper had achieved the purpose of identifying Zanthoxylurn Bungeanum species non-destructively,rapidly and accurately,which provided a new way for agricultural products nondestructive testing research.更多还原