Hi all. I have experience in about 1 class with Matlab but I'm stumped right now. I know I could do this manually, but I'm trying to practice with Matlab. My problem is that stress-strain data is usually approximately linear up to a point, then it becomes non-linear. Is there a way to "extract" the linear portion of data and put it in a matrix? Here's some example data: clear; clc; close all; A= [0 0 331 0.156 669 0.315 1022 0.493 1140 0.554 1175 0.581 1167 0.776 1177 1.294 1193 2.16 1207 2.98 1217 3.85 1221 4.77 1200 6.02 1129 7.51 1045 9.03 944 10.60 829 12.00]; stress = A(:,1); strain = A(:,2)/100; plot(strain,stress) I was thinking about trying to do something like "find the first place where the gradient switches from positive to negative"... but my real-world data isn't strictly increasing since there's errors in measurements. I almost feel like I'm thinking too hard about it and there might be a completely different way that's more efficient... Anyone have an idea?

Extract linear region of data (stress-strain data)

6 vues (au cours des 30 derniers jours)

Michael le 1 Oct 2015

0
Lien

Utiliser le lien direct vers cette question

https://fr.mathworks.com/matlabcentral/answers/246157-extract-linear-region-of-data-stress-strain-data

Modifié(e) : Michael le 1 Oct 2015

Hi all. I have experience in about 1 class with Matlab but I'm stumped right now. I know I could do this manually, but I'm trying to practice with Matlab.

My problem is that stress-strain data is usually approximately linear up to a point, then it becomes non-linear. Is there a way to "extract" the linear portion of data and put it in a matrix?

Here's some example data:

   clear; clc; close all;
     A= [0 0
0.156
0.315
0.493
0.554
0.581
0.776
1.294
2.16
2.98
3.85
4.77
6.02
7.51
9.03
10.60
12.00];
 stress = A(:,1); 
 strain = A(:,2)/100; 
 plot(strain,stress)

I was thinking about trying to do something like "find the first place where the gradient switches from positive to negative"... but my real-world data isn't strictly increasing since there's errors in measurements.

I almost feel like I'm thinking too hard about it and there might be a completely different way that's more efficient...

Anyone have an idea?