how to write a discrete function for grey-box

Hi there...

i have a model regressory like below that i want to put it in my "m-file.m" as a function of idnlgrey command.

   midnlgrey = idnlgrey(m-file.m,de)

my regresory model is :

y1(t) = 0.33*u2(t)*x1(t-1)^(1.34)+1.82*u1(t)+-0.01*u3(t)*u2(t)+1.3*x2(t-2);
y2(t) = 0.6*u2(t)-0.16*x1(t-1)^(9/8)+x2(t-1);
y3(t) = 1.65*u3(t)+3*u2(t-1)^2+u3(t-2)+7*u2(t-2);

how can i write above regressory equations as discrete function for idnlgrey command.

thank you

Answer 1

[Walter Roberson]

   midnlgrey = idnlgrey('m_file', de)

You do not need to specify the '.m' extension

Notice that I renamed m-file.m to m_file.m . The name needs to be the name of a function and "-" is not permitted in the name of MATLAB functions.

http://www.mathworks.com/help/ident/ug/estimating-nonlinear-grey-box-models.html

The file m_file.m would have to start with

function [dx, y] = m_file(t,x,u,p1,p2, ...,pN)

where the number of p* arguments depends upon the number of parameters you need.

I cannot tell at the moment which of your u1, u2, u3, x1, x2, x3 are intended to be inputs, and which are intended to be states, and which are intended to be functions so I will not suggest the code for your m_file.m at this time.

[MJ]

Thanks walter, but i know the form of "idnlgrey" command.as u know,"m-file' defined in differential or difference equations, but i want to define regressory nonlinear model in "m-file". i dont know how do it...?