1 简介
Vibration signal analysis has become one of the important methods for machinery fault diagnosis. Extraction of weak fault features from vibration signals with heavy background noise remains a challenging problem. In this paper, we first introduce the idea of algorithm-aware sparsity-assisted methods for fault feature enhancement, which extends model-aware sparsity-assisted fault diagnosis and allows a more flexible and convenient algorithm design. In the framework of algorithm-aware methods, we define the generalized structured shrinkage operators and construct the generalized structured shrinkage algorithm (GSSA) to overcome the disadvantages of l1-norm regularization based fault feature enhancement methods. We then perform a series of simulation studies and two experimental cases to verify the effectiveness of the proposed method. Additionally, comparisons with model-aware methods, including basis pursuit denoising and windowed-group-lasso, and fast kurtogram further verify the advantages of GSSA for weak fault feature enhancement.
2 部分代码
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%
clc
clear all
close all
addpath(genpath(fileparts(mfilename('fullpath'))));
rng('default')
rng(19)
%% Figure initialization
global FontSize FontName;
Tstring = 'Time (s)';
Fstring = 'Frequency (Hz)';
Astring = 'Amp (m/s^2)';
FontSize = 11; FontName = 'Times New Roman';
MarkerSize = 4; LineWidth = 1;
%%
FlagFigureAutoSave = 1;
currentFolder = pwd;
%% Simulation
Fs = 20480;
N = 4096;
mode = 'outer';
% [Sig_Impulse , t] = MakeSignalBearing( Fs, N, 'outer');
Sig_Impulse = QuasiPeiodicImpulseResponse_AM(N, Fs);
t = (0 : N-1) / Fs;
Sig_Cos = 0.5 * cos(2*pi*160*t') + 0.3 * cos(2*pi*320*t');
Sigma = 0.6;
Noise = Sigma * randn(N , 1);
Sig_Combine = Sig_Cos + Sig_Impulse' + Noise;
% Sig_Combine = Sig_Cos + Sig_Impulse + Noise;
[ yf2, f2 ] = Dofft( Sig_Combine , Fs , 0);
K = 10 : 10 : 1000;
for i = 1 : length(K)
%% Setting the parameters
i
Q = 2;
r = 5;
J =10;
now = ComputeNow(N,Q,r,J,'radix2');
AH = @(Sig) tqwt_radix2(Sig, Q, r, J);
A = @(w) itqwt_radix2(w, Q, r , N);
lam = 1.0 * now;
rho = 1;
%% method1 : Generalized Structured Shrinkage
K1 = K(i);
Method1.Name = 'WGL';
Method1.Initial_Size = 5;
Method1.SubName = 'MC';
Method1.gamma = 2;
Method1.window = 'gausswin';
z1 = IterGSS(Sig_Combine, A, AH, lam, rho, K1, Method1);
%% method2 : TQWT-L1 based
Method2.Name = 'L1';
K2 = K(i);
z2 = IterGSS(Sig_Combine, A, AH, lam, rho, K2, Method2);
%% method3 : Neighbor thresholding
params.Q = 2;
params.r = 5;
params.J =10;
K3 = K(i);
z3 = TQWTDe( Sig_Combine, params , 'nc', K3);
%% method4 : Structured Shrinkage
Method3.Name = 'WGL';
Method3.Initial_Size = 5;
Method3.SubName = 'L1';
Method3.window = 'gausswin';
K4 = K(i);
z4 = IterGSS(Sig_Combine, A, AH, lam, rho, K4, Method3);
z1 = real(A(z1));
z2 = real(A(z2));
% z3 = real(A(z3));
z4 = real(A(z4));
%% Calculate RMSE
GST_RMSE(i) = RMSE(z1, Sig_Impulse);
L1_RMSE(i) = RMSE(z2, Sig_Impulse);
NC_RMSE(i) = RMSE(z3', Sig_Impulse);
ST_RMSE(i) = RMSE(z4, Sig_Impulse);
end
save('Harmonic_Inference_Best_K.mat', 'K', 'GST_RMSE', 'L1_RMSE', 'NC_RMSE', 'ST_RMSE')
3 仿真结果
4 参考文献
[1] Zhao Z , Wang S , Xu W , et al. Sparsity-Assisted Fault Feature Enhancement: Algorithm-Aware Versus Model-Aware[J]. IEEE Transactions on Instrumentation and Measurement, 2020, PP(99):1-1.
博主简介:擅长智能优化算法、神经网络预测、信号处理、元胞自动机、图像处理、路径规划、无人机等多种领域的Matlab仿真,相关matlab代码问题可私信交流。
部分理论引用网络文献,若有侵权联系博主删除。