基于红外光谱的白酒鉴伪及溯源系统设计与实现

发布时间：2018-01-20 00:15

本文关键词： 红外光谱特征提取鉴伪溯源　出处：《华中科技大学》2015年硕士论文　论文类型：学位论文

【摘要】：白酒是我国的特有酒类,历史悠久。近些年来,随着白酒需求量的提升,大量的假酒涌入市场,这些假酒不但危害消费者的身体健康,损害消费者的利益,同时也使酒厂经济损失严重,扰乱了正常的经济秩序。因此,如何快速、有效地对白酒进行真伪鉴别和假酒溯源成为公安机关等执法部门关注的重点问题。红外光谱分析技术是最近几十年兴起的一门光谱分析技术,吸引着大量的学者投入研究。红外光谱分析法具有测定迅速、无损分析、信息丰富等优点,其为白酒检测引入了一个新思路。本文以采集到的真假酒红外光谱为研究对象,重点研究了白酒红外光谱特征提取技术、白酒红外光谱分类技术以及白酒红外光谱聚类技术,旨在设计和开发一种能够对白酒红外光谱进行数据存储、特征提取,对白酒进行真伪鉴别以及假酒溯源的软件系统,以协助监管部门打击白酒的制假贩假,主要内容与研究结果如下:(1)总结了传统的红外光谱特征提取方法及其优缺点,提出了一种基于小波的多频段特征提取方法。实验结果表明,提取方法能有效提取白酒红外光谱在各频段上有重要区分作用的特征点,提高分类模型的预测准确率。(2)将支持向量机与马氏距离判别法结合起来对采集到的白酒真酒光谱进行分类研究,提出一种基于白酒属性特征的层次化白酒多分类方案。实验结果表明,采用多频段特征提取方法结合层次分类方案建立的分类模型预测正确率最高,达到了91.14%。(3)采用主成分分析与DBSCAN算法相结合的方法对采集到的假酒红外光谱数据进行聚类研究。实验结果表明,聚类可以作为一种有效的假酒窝点和勾兑配方的溯源方法。(4)综合本文所提出的白酒红外光谱特征提取方法、分类方法以及聚类方法设计了一套基于红外光谱的白酒鉴伪及溯源综合技术方案,然后基于该方案设计并实现了白酒鉴伪及溯源系统。经过测试,该系统能实现白酒真伪鉴定以及假酒溯源。
[Abstract]:Liquor is a unique liquor in China, with a long history. In recent years, with the promotion of liquor demand, a large number of counterfeit wine poured into the market, these fake liquor not only harm the health of consumers, harm the interests of consumers. At the same time, it also causes serious economic losses to the brewery and disrupts the normal economic order. Therefore, how to quickly. Effective identification of liquor authenticity and tracing the origin of counterfeit liquor has become the focus of law enforcement departments such as public security organs. Infrared spectrum analysis technology is a spectral analysis technology developed in recent decades. Infrared spectrum analysis has the advantages of rapid measurement, nondestructive analysis, rich information and so on. It introduced a new idea for the detection of liquor. In this paper, the infrared spectrum of real and fake liquor was taken as the research object, and the extraction technology of infrared spectrum characteristics of liquor was studied emphatically. The classification technology of infrared spectrum of liquor and the clustering technology of infrared spectrum of liquor are designed and developed in order to design and develop a kind of data storage and feature extraction for infrared spectrum of liquor. The software system to identify the authenticity of liquor and trace the origin of the fake liquor to help the regulatory authorities to crack down on the fake liquor counterfeiting. The main contents and results are as follows: (1) the traditional infrared spectrum feature extraction method and its advantages and disadvantages are summarized, and a wavelet based multi-band feature extraction method is proposed. The experimental results show that. Extraction method can effectively extract the infrared spectrum of liquor in each frequency band has an important role in distinguishing feature points. Improve the prediction accuracy of classification model. 2) combine support vector machine and Markov distance discriminant to classify the spectrum of true liquor collected. A hierarchical liquor multi-classification scheme based on liquor attributes was proposed. The experimental results showed that the classification model based on multi-band feature extraction combined with hierarchical classification had the highest prediction accuracy. The method of principal component analysis (PCA) and DBSCAN algorithm was used to cluster the infrared spectrum data of fake wine. Clustering can be used as an effective method to trace the origin of the faking liquor and blending formula. 4) synthesizing the infrared spectral feature extraction method of liquor proposed in this paper. Classification method and clustering method designed a comprehensive scheme of identification and traceability of liquor based on infrared spectrum, and then designed and realized the system of identification and traceability of liquor based on this scheme. The system can be used to identify the true and false liquor and trace the origin of the fake liquor.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TS261.7;TN219

【相似文献】