# EM-for-BAYOMA
Bayesian operational modal analysis based on the expectation-maximization algorithm.
An expectation-maximization is proposed to speed up the original Bayesian FFT method for operational modal analysis.
Fast computations for both the most probable value (MPV) and the posterior covariance matrix (PCM) are invloved.
Before running the algorithm, the frequency bands should be manually selected by conbining the information in the spectra.
The significant peak in the power spectrum density (PSD) tells where the structural mode possibly locates.
The singular value (SV) spectrum consists of eigenvalues of the PSD matrix at each frequency point.
In a resonance band, the number of lines significantly above others reveals the number of modes, and other lines indicate the noise.
An example can be found as below:
clear all; close all; clc
addpath(genpath(pwd)); % setpath
%% true values
f = [0.98 1 1.02]; % frequencies, Hz
z = [0.8 1 1.2]/100; % damping ratios, 1
phi = [1 2 2;2 1 -2;1 -2 2]'/3; % mode shapes
S = blkdiag([1 0.5*exp(1i*pi/4); 0.5*exp(-1i*pi/4) 1],1); % modal force PSD, \mug^2/Hz
Se = 100; % channel noise PSD \mug^2/Hz
%% identification - bayoma
in = load('modes3.mat');
% in.tdata contains measured data;
% in.fs is the sampling frequency, Hz;
in.f0 = {[0.98 1.0 1.02]}; % initial guess of frequencies
% if there are multiple bands, in.f0 = {[f01, f02]; [f03 f04]; ...}
in.f1f2 = [0.85 1.15]; % selected frequency bands
% if there are multiple bands, in.f1f2 =[lower1 upper1; lower2 upper2; ...]
% IN contains the following mandatory fields:
% tdata = (nt,n) ambient data; nt = no. of time points; n = no. of dofs
% fs = scalar, sampling rate (Hz)
% f1f2 = (nb,2), f1f2(:,1) and f1f2(:,2) gives the lower and upper bound
% of frequency bands used for modal identification, nb = no. of
% bands
% f0 = {nb,1}, cell array of initial guesses of natural frequencies
%
% The following optional fields can be supplied in IN:
% maxiter = max. no. of iterations in determining MPV
% p = power, fft data is multiplied by (2*pi*f)^p.
% This power converts the data to acceleration data assumed in the
% theory of the program.
% E.g., p=0 for acceleration data (default)
% p=1 converts velocity data to acceleration data
% p=2 converts displacment data to acceleration data
% The measurementn error of data is assumed to have constant PSD in each
% selected band.
out = em4bayoma(in);
% OUT is a structure with the following fields:
% 1. MPV of modal parameters:
% f = (1,m) natural frequencies (Hz)
% z = (1,m) damping ratios
% phi = (n,m) mode shape, each column normalized to unity
% Se = scalar, prediction error variance
% S = (m,m) PSD matrix of modal force
%
% Note: If tdata is given in g and fs in Hz, then Se and S are two-sided
% with a unit of g^2/Hz
%
% 2. Posterior uncertainty
% coefv = cell containing blocks of c.o.v. for
% f(:),z(:),S(:),Se,phi(:,1),phi(:,2),...,phi(:,m),
% Note that off-diagonal terms of c.o.v. of S is the c.o.v. of coherence
没有合适的资源?快使用搜索试试~ 我知道了~
基于期望最大化算法的贝叶斯操作模态分析matlab代码.zip
共27个文件
m:25个
mat:1个
md:1个
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
0 下载量 117 浏览量
2024-02-21
23:22:43
上传
评论
收藏 11.18MB ZIP 举报
温馨提示
1.版本:matlab2014/2019a/2021a 2.附赠案例数据可直接运行matlab程序。 3.代码特点:参数化编程、参数可方便更改、代码编程思路清晰、注释明细。 4.适用对象:计算机,电子信息工程、数学等专业的大学生课程设计、期末大作业和毕业设计。
资源推荐
资源详情
资源评论
收起资源包目录
基于期望最大化算法的贝叶斯操作模态分析matlab代码.zip (27个子文件)
基于期望最大化算法的贝叶斯操作模态分析matlab代码
EM-for-BAYOMA-main
modes3.mat 11.14MB
Example.m 2KB
EM4BayOMA
PCM
cstgradnull.m 1KB
NLLFHess_scalarSe.m 15KB
EYitak12.m 2KB
ABterms_ScalarSe.m 6KB
support
calMAC.m 940B
com_mat.m 735B
pageprodtr.m 2KB
vec2diag.m 549B
pagemtimes2.m 2KB
dup_mat.m 2KB
vec.m 1KB
pagevec.m 2KB
pagekron.m 881B
pick_vec.m 2KB
elim_mat.m 1KB
pagediag.m 1KB
EYitak34.m 3KB
PostCOV_scalarSe.m 11KB
em4bayoma.m 6KB
MPV
P2P.m 247B
Q2.m 653B
oneloglik.m 956B
bayoma_em.m 10KB
oneEMstep.m 2KB
README.md 3KB
共 27 条
- 1
资源评论
Matlab科研辅导帮
- 粉丝: 2w+
- 资源: 7669
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功