發(fā)布時(shí)間：2018-12-15 11:50

【摘要】：在現(xiàn)代社會(huì)中，人們對(duì)食品質(zhì)量檢測(cè)的要求越來(lái)越高，而傳統(tǒng)的化學(xué)分析方式檢測(cè)時(shí)間長(zhǎng)、成本高、效率低，且需要專業(yè)人員進(jìn)行操作，不利于普及和推廣。而檢測(cè)方便且快速的電子鼻系統(tǒng)的出現(xiàn)正在代替?zhèn)鹘y(tǒng)的化學(xué)分析方式，目前，雖然電子鼻技術(shù)取得了極大的進(jìn)展，但大多數(shù)還停留在實(shí)驗(yàn)室階段，還沒(méi)有廣泛實(shí)用化，主要原因是電子鼻的漂移問(wèn)題、背景干擾、個(gè)體差異等問(wèn)題還沒(méi)有得到有效解決。 通過(guò)長(zhǎng)期實(shí)驗(yàn)數(shù)據(jù)研究分析發(fā)現(xiàn)，漂移主要分為短期漂移與長(zhǎng)期漂移兩種，其中長(zhǎng)期漂移具有變化緩慢且無(wú)規(guī)律的現(xiàn)象。本文把降低或者消除電子鼻的長(zhǎng)期漂移作為主要研究?jī)?nèi)容。 為了解決電子鼻長(zhǎng)期漂移導(dǎo)致鑒別正確率降低這一問(wèn)題，提出了一種基于快速傅里葉變換的均值偏差率閾值函數(shù)來(lái)去除電子鼻漂移噪聲的方法，通過(guò)構(gòu)造快速傅里葉變換系數(shù)的均值偏差率閾值函數(shù)，實(shí)現(xiàn)了對(duì)傅里葉變換系數(shù)的動(dòng)態(tài)處理，有效去除了電子鼻漂移噪聲。6種白酒檢測(cè)實(shí)例表明，此方法能夠有效消除電子鼻長(zhǎng)期漂移噪聲信號(hào)對(duì)傳感器造成的影響，從而顯著的提高了鑒別正確率。 傅里葉變換刻畫(huà)的是電子鼻檢測(cè)信號(hào)在整個(gè)時(shí)域的頻譜特征，雖然在去除電子鼻漂移信號(hào)過(guò)程中有較好的應(yīng)用效果，但是并沒(méi)有獲得漂移信號(hào)的特征信息。針對(duì)這一問(wèn)題，本文提出一種在小波變換的基礎(chǔ)上結(jié)合均值偏差率閾值函數(shù)對(duì)小波變換系數(shù)進(jìn)行處理的方法，由于小波變換可以實(shí)現(xiàn)電子鼻檢測(cè)信號(hào)在時(shí)域和頻域的局部變換，克服了傅里葉變換的缺點(diǎn)，便于觀察分析。實(shí)驗(yàn)結(jié)果表明，與傅里葉變換相比，小波變換能夠進(jìn)一步去除電子鼻的漂移信號(hào)，從而使分類效果更佳。 獨(dú)立成分分析能從檢測(cè)的混合信號(hào)中分離出分布未知但統(tǒng)計(jì)相互獨(dú)立的源信號(hào)，，結(jié)合漂移信號(hào)未知性等特點(diǎn)，本文嘗試從獨(dú)立成分分析的角度對(duì)傳感器的檢測(cè)信號(hào)進(jìn)行分析研究。通過(guò)利用獨(dú)立成分分析，提取與白酒揮發(fā)出的氣體最相關(guān)的獨(dú)立成分，然后對(duì)其進(jìn)行分析研究，探究漂移規(guī)律，從而達(dá)到去除漂移噪聲信號(hào)的目的。實(shí)驗(yàn)結(jié)果表明，獨(dú)立成分分析能夠有效的提取獨(dú)立成分，去除了一定的漂移信號(hào)。
[Abstract]:In modern society, the demand of food quality detection is more and more high, but the traditional chemical analysis method has long time, high cost, low efficiency, and requires professional operation, which is not conducive to popularization and promotion. The appearance of the electronic nose system, which is convenient and rapid, is replacing the traditional chemical analysis method. Although the electronic nose technology has made great progress at present, most of them are still in the laboratory stage and have not been widely applied. The main reasons are that the drift of electronic nose, background interference and individual differences have not been effectively solved. Through the analysis of long-term experimental data, it is found that the drift is mainly divided into two types: short-term drift and long-term drift, and the medium- and long-term drift has the phenomenon of slow and irregular change. In this paper, reducing or eliminating long-term drift of electronic nose is the main research content. In order to solve the problem that the long term drift of electronic nose leads to the reduction of the discrimination accuracy, a method based on the mean deviation rate threshold function based on fast Fourier transform (FFT) is proposed to remove the drift noise of electronic nose. By constructing the mean deviation rate threshold function of the fast Fourier transform coefficient, the dynamic processing of the Fourier transform coefficient is realized, and the electronic nose drift noise is effectively removed. This method can effectively eliminate the effect of the long-term drift noise signal on the sensor and improve the discrimination accuracy. Fourier transform depicts the spectrum characteristics of the electronic nose detection signal in the whole time domain. Although it has good application effect in the process of removing the electronic nose drift signal, it does not obtain the characteristic information of the drift signal. In order to solve this problem, this paper presents a method to deal with wavelet transform coefficients based on wavelet transform combined with mean deviation rate threshold function. Because wavelet transform can realize the local transformation of electronic nose detection signal in time domain and frequency domain. It overcomes the shortcoming of Fourier transform and is easy to observe and analyze. The experimental results show that the wavelet transform can remove the drift signal of the electronic nose further than the Fourier transform, so that the classification effect is better. Independent component analysis (ICA) can separate source signals with unknown distribution but independent statistics from the detected mixed signals, combining with the unknown characteristics of drift signals. In this paper, the detection signals of sensors are analyzed and studied from the angle of independent component analysis (ICA). Through the use of independent component analysis, extract the most relevant independent components of the gas volatilized from liquor, then analyze and study it, explore the drift law, so as to achieve the purpose of removing drift noise signal. The experimental results show that independent component analysis can effectively extract independent components and remove certain drift signals.
【學(xué)位授予單位】：河南科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：TP212;TB535

【參考文獻(xiàn)】

中國(guó)期刊全文數(shù)據(jù)庫(kù) 前10條

1 喬玉坤;王仕成;張金生;張琪;孫淵;;小波域軟閾值降噪的地磁匹配導(dǎo)航仿真研究[J];兵工學(xué)報(bào);2011年09期

2 王立川;張覃軼;黃偉;;基于氣體傳感器陣列的白酒特征分析[J];傳感技術(shù)學(xué)報(bào);2010年12期

3 張紅梅;王俊;余泳昌;高獻(xiàn)坤;花恒明;何玉靜;;基于電子鼻技術(shù)的信陽(yáng)毛尖茶咖啡堿檢測(cè)方法[J];傳感技術(shù)學(xué)報(bào);2011年08期

4 惠國(guó)華;吳玉玲;陳裕泉;;基于電子鼻的香蕉儲(chǔ)存時(shí)間鑒別方法研究[J];傳感技術(shù)學(xué)報(bào);2012年05期

5 高明星;王曉寧;龔煈;廖青;;電子鼻在棉織物異味檢測(cè)中的應(yīng)用研究[J];傳感器與微系統(tǒng);2011年06期

6 周海濤;殷勇;;主成分分析在電子鼻鑒別分析中存在的缺陷及其改進(jìn)方法[J];傳感器世界;2008年10期

7 盧松升;劉正之;;基于小波的電力系統(tǒng)信號(hào)預(yù)定閾值去噪技術(shù)[J];電工技術(shù)學(xué)報(bào);2005年11期

8 張青春;葉小婷;;基于電子鼻技術(shù)的混合氣體檢測(cè)方法研究[J];電子技術(shù)應(yīng)用;2011年12期

9 萬(wàn)靈達(dá);楊曉光;;一種基于FFT的高精度頻率估計(jì)算法[J];電子科技;2010年10期

10 文玉梅,李平;通過(guò)信號(hào)分離抑制離散變化變量傳感器漂移噪聲[J];電子學(xué)報(bào);1999年11期

本文編號(hào)：2380586