本文選題：移動窗口 + 相似度��； 參考：《光譜學與光譜分析》2017年10期

【摘要】：為滿足白酒品牌判別領域快速檢測的迫切需求,提出了一種以紫外光譜為輸入,移動窗口相關系數(shù)法為手段的白酒品牌快速判別新方法。研究以海之藍酒作為對象品牌,實驗結果顯示各批次海之藍酒在實驗波長區(qū)間存在較高的吻合一致性,基于紫外光譜原始譜圖的相關系數(shù)值和基于移動窗口光譜計算所得的相關系數(shù)值均達到0.99以上。由于其他洋河系列白酒與海之藍酒均屬于洋河酒廠生產(chǎn),原料來源、生產(chǎn)制備工藝等存在較高的相似性,此時若僅使用紫外光譜原始譜圖數(shù)據(jù)結合相似度算法進行品牌判別,則定量化的判別區(qū)分存在一定困難。移動窗口法可有效提高譜圖細節(jié)差異,突出光譜在特征譜段的形狀、強度、變化趨勢差異,提高實驗譜圖在細節(jié)結構差異辨認分析能力。經(jīng)過研究篩選,發(fā)現(xiàn)以移動窗口相關系數(shù)結合紫外光譜特征譜段(270~295nm)可提高洋河系列酒間的譜圖差異,實現(xiàn)這種有較高相似性白酒品牌的判別區(qū)分。此外,市售六種其他品牌白酒及乙醇作為對照組,進一步論證了該方案的可行性。該研究的創(chuàng)新之處在于提出了一種移動窗口相關系數(shù)光譜法進行白酒品牌快速判別分析的新思路,該方法同時具備速度快,操作簡便等優(yōu)點,具有較高的實際應用潛在價值,并可為其他食品的質(zhì)量安全判別分析研究提供借鑒。
[Abstract]:In order to meet the urgent need of rapid detection in the field of liquor brand identification, a new method of rapid identification of liquor brand was proposed, which was based on UV spectrum and moving window correlation coefficient method. The experimental results show that there is a high consistency in the experimental wavelength range of each batch of Haizhi-blue wine. The correlation coefficients based on the original UV spectrum and the moving window spectrum are both above 0.99. Because other Yanghe series liquor and Haizhi-blue liquor belong to Yanghe distillery, there is a high similarity in raw material source and production preparation technology. At this time, if only the original spectrum data of ultraviolet spectrum combined with similarity algorithm are used to distinguish the brand, There are some difficulties in quantitative discrimination. The moving window method can effectively improve the detail difference of spectrum, highlight the difference of shape, intensity and variation trend of spectrum in characteristic spectrum, and improve the ability of distinguishing and analyzing the difference of experimental spectrum in detail structure. It was found that the correlation coefficient of moving window combined with the characteristic spectrum of UV spectrum could improve the spectral difference among Yanghe series liquor and realize the discrimination of this liquor brand with high similarity. In addition, six other brands of liquor and ethanol on the market as control group, further demonstrated the feasibility of the scheme. The innovation of this study is that a new idea of fast discriminant analysis of liquor brands by moving window correlation coefficient spectrum method is put forward. This method has the advantages of high speed and easy operation, and has high potential value in practical application. It can be used for reference for other food quality and safety discriminant analysis.
【作者單位】： 南京財經(jīng)大學管理科學與工程學院;江蘇省質(zhì)量安全工程研究院;西南交通大學數(shù)學學院;浙江工商大學管理工程與電子商務學院;
【基金】：江蘇省高校自然科學研究面上項目(16KJB150015) 國家自然科學基金項目(61602217,71373117,71433006)資助
【分類號】：O657.3;TS262.3

【相似文獻】

相關碩士學位論文 前1條

1 馬凱升;FTNIR光譜結合化學計量學用于白酒品牌鑒別的快速判別分析[D];暨南大學;2016年

，

本文編號：1809468