How to use the polyfit block in Simulink
13 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
I want to do the same thing in simulink as the commands shown below: clear all clc x=[0:0.01:1]'; a=[-2 5 6]'; f=@(x) a(1)*x.^2+a(2)*x+a(3); y=f(x); factor=polyfit(x,y,2); factor =
-2.0000 5.0000 6.0000
But when I use the "simin" that read the [x,y] as the input and polyfit block to assess the coefficients, and I set the Control points(X) with x, there are two error alerts:
1)Error in port widths or dimensions. Output port 1 of 'mean_0525/Least Squares Polynomial Fit/Constant' is a [101x3] matrix.
2)Error reported by S-function 'sdspqreslv2' in 'mean_0525/Least Squares Polynomial Fit/QR Solver':
The number of rows of input A must match the number of rows of input B.
How could I solve this problem. Thanks!
0 comentarios
Respuestas (6)
Kaustubha Govind
el 26 de Mayo de 2011
How are you feeding in 'y' to the "Least Squares Polynomial Fit" block? The length of the input to the block must be the same as the length of 'x'. Try using a "From Workspace" block to get 'y'.
Fangjun Jiang
el 26 de Mayo de 2011
A wild guess. Can you try the Control points as x'? since your x is 101x1.
Arnaud Miege
el 26 de Mayo de 2011
5 comentarios
Arnaud Miege
el 26 de Mayo de 2011
If your a and x are changing at each time step, then you probably need to use a MATLAB Function block (http://www.mathworks.com/help/releases/R2011a/toolbox/simulink/slref/matlabfunction.html). I tried it with the following code:
function [y, factor] = fcn(a,x)
%#codegen
y = a(1)*x.^2+a(2)*x+a(3);
factor = polyfit(x,y,2);
and it works well. I fed to the x input a ramp signal with a slope of 1e-3 so that it ramps from 0 to 1 in 10s, and to the a input a constant block with [-2 5 6].
Note that this is in R2011a. In previous releases, the MATLAB Function block was called "Embedded MATLAB Function" block and the code may be slightly different, in particular the %#codegen hasthtag.
HTH,
Arnaud
Arnaud Miege
el 26 de Mayo de 2011
If a doesn't change, then surely factor = a? If that's the case, a simple Polynomial block parameterized with a should do the trick and produce y when fed with x. There's no need for polyfit.
If a does change, use the Embedded MATLAB Function block to include the MATLAB code in the Simulink model, as suggested in my previous comment.
0 comentarios
Fangjun Jiang
el 27 de Mayo de 2011
I am kind of lost. I don't have the DSP toolbox. Can you repeat the example shown here in Polyfit? Try this before using the Inport block to get in your input.
0 comentarios
Ver también
Categorías
Más información sobre Sources en Help Center y File Exchange.
Productos
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!