Clear Filters
Clear Filters

i am not getting the exact data for finite element model updating

11 views (last 30 days)
clear;
close all;
clc;
% __________ M3=40tons
% | | K3=50*10^6
% ---------- M2=45tons
% | | K2=50*10^6
% ---------- M1=50tons
% | | K1=50*10^6
% --- ---
set-up for stiffness matrix and other
m1=50000;
m2=50000;
m3=50000;
mass=diag([m1 m2 m3]);
fprintf('The mass matrix \n');
disp(mass)
k1=50*10^6;
k2=50*10^6;
k3=50*10^6;
stiffness=[k1+k2 -k2 0;-k2 k2+k3 -k3;0 -k3 k3];
fprintf('The stiffness matrix \n');
disp(stiffness)
Natural frequency(eigen value) and Mode shape(eigen vector) calculation
[v,d]=eig(stiffness,mass);
fprintf('lambda value \n');
disp(d);
w=[sqrt(d)];
w1=w(1,1);
w2=w(2,2);
w3=w(3,3);
W=[w1,w2,w3];
fprintf('The natural frequencies of this structure are as follows(HZ) \n');
disp(W)
fprintf('The mode shape of the structure \n');
disp(v)
fprintf('lambda value \n');
disp(d);
w=[sqrt(d)];
w1=w(1,1);
w2=w(2,2);
w3=w(3,3);
W=[w1,w2,w3];
fprintf('The natural frequencies of this structure are as follows(HZ) \n');
disp(W)
fprintf('The mode shape of the structure \n');
disp(v)
Individusal mode shapes vector
fprintf('individusal mode shapes \n');
v1=v(:,1)
v2=v(:,2)
v3=v(:,3)
Plotting of mode shape
% height=[0;3;6;9];
% for i=1:5
% subplot(1,3,i)
% plot([0; v(:,i)],height);
% xlabel('mode shape','FontSize',12);
% ylabel('Height of the structure','FontSize',12);
% title(['mode shape', num2str(i)],'FontSize',18);
% end
Sensitivity matrix
l1 = [1 0 0;0 0 0;0 0 0]; % DIFFERENTIATION OF STIFFNESS WRT K1
l2 = [1 -1 0;-1 1 0;0 0 0]; % DIFFERENTIATION OF STIFFNESS WRT K2
l3 = [0 0 0;0 1 -1;0 -1 1]; % DIFFERENTIATION OF STIFFNESS WRT K3
Mode shape decomposition
a111=0;
a112=(v2'*l1*v1)/(w1-w2);
a113=(v3'*l1*v1)/(w1-w3);
a121=0;
a122=(v2'*l2*v1)/(w1-w2);
a123=(v3'*l2*v1)/(w1-w3);
a131=0;
a132=(v2'*l3*v1)/(w1-w2);
a133=(v3'*l3*v1)/(w1-w3);
a211=(v1'*l1*v2)/(w2-w1);
a212=0;
a213=(v3'*l1*v2)/(w2-w3);
a221=(v1'*l2*v2)/(w2-w1);
a222=0;
a223=(v3'*l2*v2)/(w2-w3);
a231=(v1'*l3*v2)/(w2-w1);
a232=0;
a233=(v3'*l3*v2)/(w2-w3);
a311=(v1'*l1*v3)/(w3-w1);
a312=(v2'*l1*v3)/(w3-w2);
a313=0;
a321=(v1'*l2*v3)/(w3-w1);
a322=(v2'*l2*v3)/(w3-w2);
a323=0;
a331=(v1'*l3*v3)/(w3-w1);
a332=(v2'*l3*v3)/(w3-w2);
a333=0;
SENSITIVITY MATRIX FORMATION (CONSIDERING BOTH EIGEN VALUE AND MODE-SHAPE DECOMPOSITION)
s11 = v1'*l1*v1;
s12 = v1'*l2*v1;
s13 = v1'*l3*v1;
s21=a111*v1+a112*v2+a113*v3;
s22=a121*v1+a122*v2+a123*v3;
s23=a131*v1+a132*v2+a133*v3;
s31 = v2'*l1*v2;
s32 = v2'*l2*v2;
s33 = v2'*l3*v2;
s41=a211*v1+a212*v2+a213*v3;
s42=a221*v1+a222*v2+a223*v3;
s43=a231*v1+a232*v2+a233*v3;
s51 = v3'*l1*v3;
s52 = v3'*l2*v3;
s53 = v3'*l3*v3;
s61=a311*v1+a312*v2+a313*v3;
s62=a321*v1+a322*v2+a323*v3;
s63=a331*v1+a332*v2+a333*v3;
S = [s11 s12 s13;s21 s22 s23;s31 s32 s33;s41 s42 s43;s51 s52 s53;s61 s62 s63]
SENSITIVITY MATRIX FORMATION 2 (CONSIDERING ONLY EIGEN VALUE DECOMPOSITION)
% s11 = v1'*l1*v1;
% s12 = v1'*l2*v1;
% s13 = v1'*l3*v1;
% s21 = v2'*l1*v2;
% s22 = v2'*l2*v2;
% s23 = v2'*l3*v2;
% s31 = v3'*l1*v3;
% s32 = v3'*l2*v3;
% s33 = v3'*l3*v3;
% S=[s11 s12 s13;s21 s22 s23;s31 s32 s33];
UPDATING PARAMETER (CONSIDERING BOTH EIGEN VALUE AND MODE-SHAPE DECOMPOSITION)
K = [k1 k2 k3];
WA = [w1 v1' w2 v2' w3 v3'];
WM = [14.7257 -0.0015 -0.0027 -0.0038 35.4870 0.0029 0.0021 -0.0031 55.2413 -0.0031 0.0033 -0.0011];
K1 = K'+inv(S'*S)*S'*(WM'-WA')
UPDATING PARAMETER 2 (CONSIDERING ONLY EIGEN VALUE DECOMPOSITION)
% K = [k1 k2 k3];
% WA = [w1 w2 w3];
% WM = [13.3221 36.7237 51.7097];
% K1 = K'+inv(S*S')*S'*(WM'-WA')
  2 Comments
Steven Lord
Steven Lord on 29 Mar 2022
It's not clear what problem you're experiencing or what question you'd like help with.
Susanta  Barik
Susanta Barik on 29 Mar 2022
I AM USING SENSITIVITY METHOD TO UPDATE FINITE ELEMENT DATA BUT WHAT THE RESULT I AM GETTING THAT IS NOT EXACT DATA . ALSO WHEN I AM COMPAIRING MY EXPERIMENTAL DATA WITH UPDATED DATA I AM GETTING NOT EXACT DATA

Sign in to comment.

Answers (1)

Nipun
Nipun on 30 May 2024
Hi Susanta,
I understand that you are facing issues with finite element model updating and not getting the exact data. Here’s a concise approach to check and improve your FEM updating process:
1. Verify Model and Data Consistency: Ensure that the mesh, material properties, boundary conditions, and loads are accurately defined.
2. Check for Convergence: Ensure your solver settings are appropriate for achieving convergence.
3. Use Optimization Toolbox: Apply optimization techniques to update your model parameters.
Heres a basic outline for using optimization in MATLAB:
% Define your objective function (difference between model and experimental data)
objectiveFunction = @(params) computeDifference(params, experimentalData);
% Set initial guess for parameters
initialParams = [E, density, poisson]; % Example parameters like Young's modulus, density, Poisson's ratio
% Define optimization options
options = optimoptions('fmincon', 'Display', 'iter', 'Algorithm', 'sqp');
% Run optimization
[optimizedParams, fval] = fmincon(objectiveFunction, initialParams, [], [], [], [], lb, ub, [], options);
disp(optimizedParams);
Replace `computeDifference` with your function to calculate the difference between model and experimental data, and define `experimentalData`, `initialParams`, `lb`, and `ub` appropriately.
Hope this helps.
Regards,
Nipun

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!