本文關(guān)鍵詞： 泄漏孔徑識(shí)別 總體局域均值分解(ELMD) KL散度 稀疏表示分類器 過完備字典　出處：《中國(guó)機(jī)械工程》2017年10期 　論文類型：期刊論文

【摘要】：針對(duì)天然氣管道泄漏受孔徑、傳感器距離、管道內(nèi)壓力等多種因素影響,特征提取及識(shí)別算法較為復(fù)雜的問題,提出了基于總體局域均值分解-相對(duì)熵的特征提取算法并結(jié)合稀疏表示分類的泄漏孔徑識(shí)別新方法。該方法采用總體局域均值分解方法對(duì)泄漏信號(hào)進(jìn)行自適應(yīng)分解,得到不同孔徑泄漏信號(hào)的特征信息,并根據(jù)KL散度選擇包含主要泄漏信息的PF分量,在此基礎(chǔ)上提取多種時(shí)頻特征參數(shù),獲取全面準(zhǔn)確表征泄漏信號(hào)的特征向量;針對(duì)小樣本復(fù)雜信號(hào)的分類,提出稀疏表示分類器實(shí)現(xiàn)泄漏孔徑準(zhǔn)確分類。該分類器采用過完備字典求得測(cè)試信號(hào)的最稀疏解,并以此解作為測(cè)試信號(hào)的稀疏重構(gòu)系數(shù),以獲取測(cè)試信號(hào)在不同類別中的重構(gòu)信號(hào),最終通過判斷測(cè)試信號(hào)與重構(gòu)信號(hào)的殘差值大小完成泄漏孔徑分類。實(shí)驗(yàn)結(jié)果表明,所提出的算法比傳統(tǒng)的SVM及BP分類算法識(shí)別準(zhǔn)確率高。
[Abstract]:The gas pipeline leakage is affected by many factors, such as aperture, sensor distance, pipeline pressure and so on, so the algorithm of feature extraction and identification is more complex. In this paper, a new method of leak aperture identification based on local mean decomposition and relative entropy is proposed, which combines with sparse representation classification. The method uses the local mean decomposition method to self-adaptively divide the leakage signal. Solution. The characteristic information of different aperture leakage signal is obtained, and the PF component which contains the main leakage information is selected according to the KL divergence. On this basis, a variety of time-frequency characteristic parameters are extracted. Obtain the characteristic vector which can represent the leakage signal completely and accurately; For the classification of small sample complex signals, a sparse representation classifier is proposed to realize accurate classification of leak aperture, which uses an overcomplete dictionary to obtain the most sparse solution of the test signal. The solution is used as the sparse reconstruction coefficient of the test signal to obtain the reconstructed signal of the test signal in different classes. Finally, the leak aperture classification is completed by judging the residual value of the test signal and the reconstructed signal. The experimental results show that the proposed algorithm is more accurate than the traditional SVM and BP classification algorithms.
【作者單位】： 燕山大學(xué)信息科學(xué)與工程學(xué)院;燕山大學(xué)河北省測(cè)試計(jì)量技術(shù)及儀器重點(diǎn)實(shí)驗(yàn)室;中國(guó)石油天然氣管道通信電力工程有限公司;
【基金】：國(guó)家自然科學(xué)基金資助項(xiàng)目(51204145) 河北省自然科學(xué)基金資助項(xiàng)目(E2013203300,E2016203223)
【分類號(hào)】：TE973.6
【正文快照】： 0引言天然氣管道泄漏會(huì)造成嚴(yán)重后果,微小泄漏是燃?xì)夤艿腊l(fā)生燃爆的主要誘因,泄漏孔徑的不同直接與危險(xiǎn)程度相關(guān),當(dāng)傳感系統(tǒng)檢測(cè)到管道發(fā)生泄漏后,盡快估計(jì)出不同泄漏孔徑,是快速制定管道搶修計(jì)劃、評(píng)估泄漏尺度的重要基礎(chǔ),對(duì)燃?xì)夤艿赖男孤┘白R(shí)別具有重要意義。天然氣管道泄

【相似文獻(xiàn)】

相關(guān)期刊論文 前3條

1 程文波;王華軍;;信號(hào)稀疏表示的研究及應(yīng)用[J];西南石油大學(xué)學(xué)報(bào)(自然科學(xué)版);2008年05期

2 蔣璐璐;駱美富;張瑜;余心杰;孔汶汶;劉飛;;汽車自動(dòng)變速箱油的近紅外光譜識(shí)別研究[J];光譜學(xué)與光譜分析;2014年01期

3 ;[J];;年期

，

本文編號(hào)：1448311