荧光光谱结合概率神经网络用于无醇啤酒的识别  

Recognition of alcohol-free beer by fluorescent spectroscopy and probabilistic neural network

作　　者：魏柏林[1] 陈国庆[1] 徐建才[1] 闫冠峰[1] 马超群[1] 朱拓[1] 高淑梅[1] 

机构地区：[1]江南大学理学院,无锡214122

出　　处：《激光技术》2010年第6期794-797,共4页Laser Technology

基　　金：江苏省高校科研成果产业化推进工程资助项目(JH08-18);江苏省研究生培养创新工程资助项目(CX08B-088Z);江苏省自然科学基金资助项目(BK2009066);教育部高校博士学科点专项科研基金资助项目(200802950005)

摘　　要：为了快速、准确地识别无醇啤酒和普通啤酒,采用荧光光谱结合概率神经网络的方法,建立了识别无醇啤酒的模型。实验中发现无醇啤酒和普通啤酒在紫外-可见光激发下,都能产生较强荧光,测得无醇啤酒荧光峰在420nm～620nm之间,荧光峰值波长为490nm左右。将小波变换处理荧光光谱得到的低频系数作为网络数据,训练、建立了概率神经网络,并对60个啤酒样本进行了识别,识别率达到了98.33%。该研究结果为无醇啤酒和普通啤酒识别提供了一种新方法。In order to identify alcohol-free beer and ordinary beer quickly and accurately, a recognition model of the alcohol-free beer was established, which was based on fluorescent spectroscopy and probabilistic neural network. It was found experimentally that both alcohol-free beer and ordinary beer excited by ultraviolet-visible light could generate strong fluorescence. The fluorescent spectrum for alcohol-free beer is within a range from 420nm to 620nm, its peak wavelength of the fluorescence is about 490nm. The approximate coefficients, obtained by wavelet transform, were used as the network data, and a probabilistic neural network was trained and constructed. The trained probabilistic neural network was employed to recognize sixty beer samples, and the recognition rate was up to 98.33%. The whole research outcomes will provide a new method for recognizing alcohol-free beer.

关 键 词：光谱学 无醇啤酒 小波变换 概率神经网络 分类识别 

分 类 号：O657.3[理学—分析化学]