基于多小波分析的滾動軸承故障診斷方法研究
發(fā)布時間：2018-08-04 10:12
【摘要】：多小波又稱向量小波,是在傳統(tǒng)小波分析的基礎(chǔ)上發(fā)展起來的一種新的小波構(gòu)造理論。它是由多個小波基函數(shù)張成的函數(shù)空間,能夠同時具備對稱性、正交性、緊支撐性和高階消失矩等特性,因而在模式識別、信號除噪方面比單小波分析更具優(yōu)勢。本文將多小波分析引入滾動軸承的故障診斷,研究內(nèi)容如下: (1)為了使一維振動信號與多小波的r維空間相匹配,本文首先圍繞信號的預(yù)處理方法展開討論。以GHM多小波和CL多小波為研究對象,以信號重構(gòu)誤差和低頻平均能量比為評價標(biāo)準(zhǔn),分別針對仿真信號和實(shí)測振動信號,確立了最佳的多小波函數(shù)及相應(yīng)的預(yù)處理方法。 (2)基于GHM多小波及系數(shù)重復(fù)行的預(yù)處理方式,研究了故障信號的降噪算法。為了避免閾值選擇對降噪效果的干擾,本文分別將自適應(yīng)閾值和奇異值分解技術(shù)與多小波分析相結(jié)合。通過軸承故障的仿真信號、實(shí)驗(yàn)信號以及工程數(shù)據(jù)的分析驗(yàn)證,表明多小波的降噪效果明顯優(yōu)于單一小波,而且更易于識別滾動軸承的早期故障特征。 (3)為了充分利用多小波分解后的各子帶信息,本文研究了基于多小波包的故障診斷方法。以Shannon熵為代價函數(shù),完成了對多小波包最優(yōu)基的搜索和識別。在此基礎(chǔ)上,將其與自適應(yīng)閾值和奇異值分解的降噪方法相結(jié)合,利用不同故障模式(改變點(diǎn)蝕大小)下的軸承振動信號驗(yàn)證了降噪方法的有效性。 (4)針對正常軸承以及0.5-5mm的內(nèi)、外圈點(diǎn)蝕共9類分析模式,構(gòu)造了故障程度識別因子。其中,以多小波包分解和譜峭度分析作為特征提取的主要方法,進(jìn)而結(jié)合特征系數(shù)的復(fù)雜度計(jì)算,完成了9種狀態(tài)的模式識別,從中可以發(fā)現(xiàn)內(nèi)、外圈故障不同的演變趨勢。 (5)基于理論研究,利用LabVIEW與Matlab聯(lián)合開發(fā)了一套融合多小波分析方法的故障監(jiān)測與診斷系統(tǒng),通過6307滾動軸承的模擬故障實(shí)驗(yàn),驗(yàn)證了該系統(tǒng)的實(shí)用性及可靠性。
[Abstract]:Multi-wavelet, also called vector wavelet, is a new wavelet construction theory developed on the basis of traditional wavelet analysis. It is a function space of multiple wavelet basis functions (Zhang Cheng), which can simultaneously possess the properties of symmetry, orthogonality, compactness and high order vanishing moments, so it has more advantages in pattern recognition and signal denoising than single wavelet analysis. In this paper, multi-wavelet analysis is introduced into the fault diagnosis of rolling bearings. The research contents are as follows: (1) in order to match the one-dimensional vibration signal with the r-dimensional space of multi-wavelet, the preprocessing method of the signal is discussed in this paper. Taking GHM multiwavelets and CL multiwavelets as research objects and taking signal reconstruction error and low frequency mean energy ratio as evaluation criteria, the simulation signals and the measured vibration signals are respectively studied. The optimal multi-wavelet function and the corresponding preprocessing method are established. (2) based on the GHM multi-small sweep coefficient repeated row preprocessing method, the noise reduction algorithm of fault signal is studied. In order to avoid the interference of threshold selection to the noise reduction effect, the adaptive threshold and singular value decomposition (SVD) techniques are combined with multiwavelet analysis in this paper. Through the analysis of simulation signal, experimental signal and engineering data of bearing fault, it is shown that the denoising effect of multi-wavelet is better than that of single wavelet. Moreover, it is easier to identify the early fault features of rolling bearings. (3) in order to make full use of the information of each sub-band after multi-wavelet decomposition, a fault diagnosis method based on multi-wavelet packet is studied in this paper. With Shannon entropy as the cost function, the search and recognition of the optimal basis of multi-wavelet packets are completed. On this basis, it is combined with adaptive threshold and singular value decomposition to reduce noise. The vibration signals of bearing under different fault modes (changing the size of pitting) are used to verify the effectiveness of the noise reduction method. (4) for the normal bearing and 0.5-5mm, there are 9 kinds of analysis modes of external ring pitting. The fault degree identification factor is constructed. Among them, multi-wavelet packet decomposition and spectral kurtosis analysis are used as the main methods of feature extraction, and then combined with the complexity calculation of the feature coefficients, the pattern recognition of 9 states is completed. (5) based on the theoretical research, a fault monitoring and diagnosis system based on LabVIEW and Matlab is developed, which is based on the simulation of 6307 rolling bearing. The practicability and reliability of the system are verified.
【學(xué)位授予單位】：北京工業(yè)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2011
【分類號】：TH133.33;TH165.3
本文編號：2163577
[Abstract]:Multi-wavelet, also called vector wavelet, is a new wavelet construction theory developed on the basis of traditional wavelet analysis. It is a function space of multiple wavelet basis functions (Zhang Cheng), which can simultaneously possess the properties of symmetry, orthogonality, compactness and high order vanishing moments, so it has more advantages in pattern recognition and signal denoising than single wavelet analysis. In this paper, multi-wavelet analysis is introduced into the fault diagnosis of rolling bearings. The research contents are as follows: (1) in order to match the one-dimensional vibration signal with the r-dimensional space of multi-wavelet, the preprocessing method of the signal is discussed in this paper. Taking GHM multiwavelets and CL multiwavelets as research objects and taking signal reconstruction error and low frequency mean energy ratio as evaluation criteria, the simulation signals and the measured vibration signals are respectively studied. The optimal multi-wavelet function and the corresponding preprocessing method are established. (2) based on the GHM multi-small sweep coefficient repeated row preprocessing method, the noise reduction algorithm of fault signal is studied. In order to avoid the interference of threshold selection to the noise reduction effect, the adaptive threshold and singular value decomposition (SVD) techniques are combined with multiwavelet analysis in this paper. Through the analysis of simulation signal, experimental signal and engineering data of bearing fault, it is shown that the denoising effect of multi-wavelet is better than that of single wavelet. Moreover, it is easier to identify the early fault features of rolling bearings. (3) in order to make full use of the information of each sub-band after multi-wavelet decomposition, a fault diagnosis method based on multi-wavelet packet is studied in this paper. With Shannon entropy as the cost function, the search and recognition of the optimal basis of multi-wavelet packets are completed. On this basis, it is combined with adaptive threshold and singular value decomposition to reduce noise. The vibration signals of bearing under different fault modes (changing the size of pitting) are used to verify the effectiveness of the noise reduction method. (4) for the normal bearing and 0.5-5mm, there are 9 kinds of analysis modes of external ring pitting. The fault degree identification factor is constructed. Among them, multi-wavelet packet decomposition and spectral kurtosis analysis are used as the main methods of feature extraction, and then combined with the complexity calculation of the feature coefficients, the pattern recognition of 9 states is completed. (5) based on the theoretical research, a fault monitoring and diagnosis system based on LabVIEW and Matlab is developed, which is based on the simulation of 6307 rolling bearing. The practicability and reliability of the system are verified.
【學(xué)位授予單位】：北京工業(yè)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2011
【分類號】：TH133.33;TH165.3
【引證文獻(xiàn)】
相關(guān)期刊論文 前3條
1 徐千;程秀芳;侯婭品;;基于小波分析的滾動軸承故障診斷研究[J];機(jī)械工程師;2012年02期
2 王紅君;賀鵬;趙輝;岳有軍;劉明明;;多小波對風(fēng)機(jī)故障信號降噪處理的比較研究[J];化工自動化及儀表;2013年02期
3 張偉;;一種新型的旋轉(zhuǎn)機(jī)械滾動軸承故障診斷方法[J];科技創(chuàng)新導(dǎo)報(bào);2012年02期
相關(guān)碩士學(xué)位論文 前2條
1 羅琴;基于MEMS慣性傳感器的微小型航姿參考系統(tǒng)的設(shè)計(jì)與研究[D];上海交通大學(xué);2012年
2 郭永偉;基于支持向量機(jī)與遺傳算法的故障模式識別及趨勢預(yù)測方法研究[D];北京化工大學(xué);2012年
，本文編號：2163577
本文鏈接：http://www.wukwdryxk.cn/kejilunwen/jixiegongcheng/2163577.html
最近更新
教材專著
