how to estimate initial conditions for lsqcurvefit function in matlab? thank u in advance.

[x,resnorm]=lsqcurvefit(@myfun,x0,xdata,ydata);

how to set x0?? xdata(voltage) and ydata(displacement), experimental data are attached.

%%function: myfun function f=myfun(x,xdata) a=x(4)*xdata+x(5); for k=1:1:1; r(k)=x(1)*k; wo(k)=0; %%weighting function for i=1:1:length(a); a_b(k,i)=a(i)-r(k); ab(k,i)=a(i)+r(k); w(k,i)=max (a_b(k,i),min(ab(k,i),wo(k))); wo(k)=w(k,i); end pi(k)=x(2)*exp(-x(3)*r(k)); pfi(k,:)=pi(k)*w(k,:); end itpfi=sum(x(1)*pfi); f=a+itpfi;

Answer 1

[Star Strider]

Your problem is likely not your initial parameter estimates. Consider that the loop starting with:

for k=1:1:1;

will only execute once. (It will execute for k = 1 and stop.)

There may be other problems. Please format your code.

[Hemavathi]

Hi.. thank u very much for you reply. that was by mistake. else my code has " " " "k=1:1:8; " ( 8 play operators) . i am giving 0- 135V, 1 HZ Triangular input to my piezoactuator, and getting 15 micrometer displacement with 4 micrometer nonlinearity. but with the model iam not getting hysteresis curve between xdata and modelled displacement. i am getting almost linear curve. i tried all possible methods , but still not able to get hysteresis plot. plz help.. thank you..

[Star Strider]

I’m having problems interpreting your code. What is the original equation you want to fit?