✅作者简介:热爱科研的Matlab仿真开发者,修心和技术同步精进,matlab项目合作可私信。
🍎个人主页:Matlab科研工作室
🍊个人信条:格物致知。
更多Matlab完整代码及仿真定制内容点击👇
🔥 内容介绍
在工业领域中,滚动轴承是一种常见的关键设备,广泛应用于各种机械设备中。然而,由于长期运转和恶劣工作环境的影响,滚动轴承容易出现故障,给生产过程带来了严重的影响。因此,及时准确地识别滚动轴承故障并采取适当的维修措施至关重要。
近年来,基于振动信号的滚动轴承故障诊断方法备受关注。振动信号中包含了丰富的故障信息,通过对其进行分析和处理,可以有效地识别滚动轴承的故障类型和严重程度。其中,局部特征自动识别候选故障频率(CFF)是一种常用的滚动轴承故障诊断方法。
在局部特征自动识别候选故障频率方法中,SCoh(Spectral Coherence)是一个重要的概念。SCoh是一种用于描述两个信号之间相干性的指标,可以帮助我们判断信号中是否存在特定频率的故障成分。具体而言,SCoh可以通过计算信号的互谱密度和自谱密度之间的比值来得到。
基于SCoh的局部特征自动识别候选故障频率的滚动轴承故障识别算法步骤如下:
-
收集振动信号数据:首先,需要在滚动轴承上安装振动传感器,收集滚动轴承在正常运行和故障状态下的振动信号数据。这些数据将作为后续分析和诊断的基础。
-
信号预处理:对收集到的振动信号进行预处理,包括去除噪声、滤波和降采样等步骤。这些预处理操作可以提高后续算法的准确性和效率。
-
计算SCoh:使用预处理后的振动信号数据,计算信号的互谱密度和自谱密度,并通过它们的比值得到SCoh。这一步骤可以帮助我们找到信号中存在的故障频率。
-
特征提取:从计算得到的SCoh中提取出局部特征,例如局部最大值和局部最小值。这些特征可以帮助我们进一步判断信号中的故障频率。
-
候选故障频率识别:根据提取的局部特征,识别出候选的故障频率。这些候选频率可能是滚动轴承故障的预警信号,需要进一步的分析和确认。
-
故障诊断和维修:根据识别出的候选故障频率,进行故障诊断和维修。根据不同的故障类型和严重程度,采取相应的维修措施,以保证滚动轴承的正常运行。
总结而言,基于SCoh的局部特征自动识别候选故障频率是一种有效的滚动轴承故障诊断方法。通过对振动信号进行分析和处理,可以准确地识别滚动轴承的故障类型和严重程度,为维修提供科学依据。在实际应用中,我们可以根据具体情况选择合适的算法和参数,以获得最佳的故障诊断效果。
📣 部分代码
set(gca,'unit','normalized','Position',[leftleave+(j-1)*step_2 downleave+(subplotnum_1-i)*step_1 plotwidth*1 plotheight*1]);figure_FontSize = 7; set(gca,'Fontsize',figure_FontSize,'Fontname','Times New Roman');if (i-1)*2+j == 1set(gca,'ytick',[-1 0 1]); set(gca,'ylim',[-1 1]);yt = 1;ys = 0;elseif (i-1)*2+j == 2set(gca,'ytick',[-2 :2: 2]); set(gca,'ylim',[-2 2]);yt = 2;ys = 0;elseif (i-1)*2+j == 3set(gca,'ytick',[-0.8 0.4 1.6]); set(gca,'ylim',[-0.8 1.6]);yt = 1.6;ys = 0.4;elseif (i-1)*2+j == 4set(gca,'ytick',[-3 0 3]); set(gca,'ylim',[-3 3]);yt = 3; ys = 0;elseif (i-1)*2+j == 5set(gca,'ytick',[-4 0 4]); set(gca,'ylim',[-4 4]);yt = 4; ys = 0;elseif (i-1)*2+j == 6set(gca,'ytick',[0 :0.1 : 0.3]); set(gca,'ylim',[0 0.3]);yt = 0.3; ys = 0.15;endif (i-1)*2+j < 6set(gca,'xtick',[0: 0.2: 1]); set(gca,'xlim',[0 1]);xlabel('Time [s]');ylabel('Amplitude','Position',[-0.10*1 ys]);xt = -0.15*1;elseset(gca,'xtick',[0: 1: Fs/2/1000]); set(gca,'xlim',[0 Fs/2/1000]);xlabel('Frequency [kHz]');ylabel('Amplitude','Position',[-0.10*Fs/2/1000 ys]);xt = -0.15*Fs/2/1000;endtext(xt,yt,str{(i-1)*2+j},'Fontname','Times New Roman','FontSize',8,'FontWeight','bold')endend%% Spectral Coherence estimatorSig = Sigplot(:,5);% Setting parametersNw = 2^8;alpha_max = 800;opt.coh = 1;opt.lent = 8;% SCoh[Scoh,alpha,f_Scoh,STFT,~,Nv] = Fast_SC(Sig,Nw,alpha_max,Fs,opt);% the 'Fast_SC.m' is a fast and effective SC estimator and proposed by Antoni in:% J. Antoni, G. Xin, N. Hamzaoui, "Fast Computation of the Spectral Correlation",% Mechanical Systems and Signal Processing, Elsevier, 2017.% A publicly available version of the algorithm coded in Octave/Matlab has been posted at the following address:% https://fr.mathworks.com/matlabcentral/fileexchange/60561.% EESScohEES = mean(abs(Scoh(:,2:end)));%% IESCFFOgramnlevel = 5; plotMode = 0;Pre = 0.05; Pre_cff = (1-Pre/t(end))*100;[CFFOgram,Level_w1,freq_w1,I1,Bw1,Fc1,SES1,EES_CFF]=IESCFFOgram( Scoh,alpha,f_Scoh,Nw,Fs,nlevel,Pre_cff,plotMode);%%figure(2)subplotnum_1 = 3;subplotnum_2 = 2;leftleave = 0.08;upleave = 0.013;downleave = 0.08;step_1 = (1-upleave-0)/subplotnum_1;step_2 = 1/subplotnum_2;plotheight = step_1*0.66;plotwidth = step_2*0.60;str = {'(a)','(b)','(c)','(d)','(e)'};figure_FontSize = 7;set(gcf,'unit','centimeters','position',[20 10 13.5 8.5], 'color','w');subplot(3,2,1);set(gca,'unit','normalized','Position',[leftleave+0*step_2 downleave+1.14*step_1 plotwidth*1.1 plotheight*2.3]);imagesc(alpha(2:end),f_Scoh/1000,abs(Scoh(:,2:end)));axis xy;colorbar;hBar = colorbar;get(hBar, 'Position'); set(hBar, 'Position', [0.419 0.455 0.03 plotheight*2.3]);set(gca,'xtick',[0 : 200 : alpha_max]);xlim([0 alpha_max]);xlabel('Cyclic frequency \alpha [Hz]','Fontsize',figure_FontSize,'Fontname','Times New Roman');ylabel('Spectral frequency f [kHz]','Fontsize',figure_FontSize,'Position',[-0.13*alpha_max 6.0]);title('spectral coherence');set(gca,'Fontsize',figure_FontSize,'Fontname','Times New Roman');xt = -0.2*alpha_max; yt = 13.6;text(xt,yt,str{1},'Fontname','Times New Roman','FontSize',8,'FontWeight','bold');subplot(3,2,2);set(gca,'unit','normalized','Position',[leftleave+1*step_2 downleave+1.14*step_1 plotwidth*1.1 plotheight*2.3]);imagesc(freq_w1/1000,1:2*nlevel,CFFOgram);colorbar;hBar = colorbar;get(hBar, 'Position'); set(hBar, 'Position', [0.92 0.455 0.03 plotheight*2.3]);title(['@ level ',num2str(round(10*Level_w1(I1))/10),' Bw = ',num2str(round(Bw1)),' Hz fc = ',num2str(Fc1),' Hz'],'Fontsize',figure_FontSize,'Fontname','Times new roman');set(gca,'ytick',1:2*nlevel,'yticklabel',round(Level_w1*10)/10);set(gca,'xtick',[0 : 3 : Fs/2/1000]);xlim([0 Fs/2/1000]);xlabel('Frequency [kHz]','Fontsize',figure_FontSize,'Fontname','Times new roman');ylabel('Level','Fontsize',figure_FontSize,'Position',[-0.15*12.8 5.5]);set(gca,'Fontsize',figure_FontSize);set(gca,'Fontname','Times new roman');xt = -2.8; yt = -0.10; text(xt,yt,str{3},'Fontname','Times New Roman','FontSize',8,'FontWeight','bold');set(gca,'Fontsize',figure_FontSize);set(gca,'Fontname','Times new roman');subplot(3,2,5);plot(alpha(2:end),ScohEES,'b');Points = FeatruePoints(alpha(2:end),ScohEES,f_m);hold on;plot(Points(1:3,1),Points(1:3,2),'ro','MarkerFaceColor','r','MarkerSize',2);for kd = 1 : 3xts = Points(kd,1)-24; yts = Points(kd,2)+0.005;if kd == 1text(xts,yts,['1x\it{f}\it{_m}'],'Fontname','Times New Roman','FontSize',7);elsetext(xts,yts,[num2str(kd),'x\it{f}\it{_m}'],'Fontname','Times New Roman','FontSize',7);endendset(gca,'unit','normalized','Position',[leftleave+0*step_2 downleave+0*step_1 plotwidth*1.27 plotheight*1.2]);xlabel('Time (s)','Fontsize',figure_FontSize,'Fontname','Times new roman');set(gca,'xtick',[0 : 100 : alpha_max]); set(gca,'xlim',[0 alpha_max]);set(gca,'ytick',[0.05 :0.01: 0.08]); set(gca,'ylim',[0.043 0.08]);xlabel('Frequency [Hz]','Fontsize',figure_FontSize,'Fontname','Times new roman');ylabel('Amplitude','Fontsize',figure_FontSize,'Position',[-0.14*alpha_max 0.0625]);xt = -0.20*alpha_max; yt = 0.080; text(xt,yt,str{2},'Fontname','Times New Roman','FontSize',8,'FontWeight','bold')set(gca,'Fontsize',figure_FontSize,'Fontname','Times new roman');subplot(3,2, 6)x = alpha(2:end);y = EES_CFF;Points = FeatruePoints(x,y,f_m);
⛳️ 运行结果

🔗 参考文献
Yao Cheng, Shengbo Wang, Bingyan Chen, Guiming Mei, Weihua Zhang, Han Peng, Guangrong Tian, "An Improved Envelope Spectrum via Candidate Fault Frequency Optimization-gram for Bearing Fault Diagnosis", Journal of Sound and Vibration,Elsevier, 2022.
本文介绍了在工业领域中,利用振动信号和SCoh(SpectralCoherence)进行滚动轴承故障诊断的方法。通过计算信号的互谱密度和自谱密度,提取局部特征,识别候选故障频率,从而实现故障的准确识别和维修决策。
1078

被折叠的 条评论
为什么被折叠?



