Problem related to fminbnd

I am finding the minimum value of Lr in the non-linear function. but whatever value of constants I change there is no effect on the minimum value. Also the minimum value I am getting is not correct.

fun=@(Lr)(((Vin*Cr*Lm)/(((((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)+((ilm*ilm*Lm*Cc)/(((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)))*(((1-D)*Ts)-((ilm*Lm)/(((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)))))+((Lr*(((((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)+((ilm*ilm*Lm*Cc)/(((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)))*(((1-D)*Ts)-((ilm*Lm)/(((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)))))/(Lm*Vin))+((C*Lr)/Vin));

[Lr,fval] = fmincon(fun,0,10e-6)

where all other variables are constants and their resp. values are given

Answer 1

[Matt J]

You could plot the function to see how its shape varies with the parameters and to confirm whether fminbnd is giving the correct result.

[Prajakta Ghatage]

I tried to plot the function but the plot consist of only one point only one point using fminbnd command. If for different values of Lr the function is plotted then I get minimum at different value

[Matt J]

I suggest you attach a .mat file with the different sets of constant data needed to reproduce what you are seeing.

[Prajakta Ghatage]

Attaching the .mat file

[Matt J]

I can't plot the function because 'Z' is not provided.

[Prajakta Ghatage]

Z is provided just above the line where Z is mention for the first time

Vcc=(D/(1-D))*Vin

Z=Vcc

[Matt J]

Because your Lr-axis has such a small scale, you need to set a correspondingly fine TolX tolerance,

[Lr_opt,fval]=fminbnd(fun,0,10e-6,optimset('TolX',1e-10));

Now, it's fine:

fplot(fun,[0.5,4]*1e-6);
hold on; plot(Lr_opt,fval,'rx','MarkerSize',10); hold off
ylim([0.8,1.7]*1e-7)

title("Variation with Lr")
xlabel("Lr")
ylabel("Interval II")

[Prajakta Ghatage]

When I do the same changes in my code I am getting the minima to be at other value

[Prajakta Ghatage]

Can you attach the .mat file where you made the changes.

Thanks in advance

[Matt J]

Presumably you changed the constants and now fun() is a different function. But it is clear from your plot that fminbnd is doing its job. The red 'x' marks the point that fminbnd believes is the minimum, and it is clear from the plot that it found its location correctly.

[Matt J]

Can you attach the .mat file where you made the changes.

Do you mean the .m file with the code? A .mat file is not code. It is a data file, generated by the save command,

https://www.mathworks.com/help/matlab/ref/save.html

Here is the complete code that I used:

clear

%% given specifications
D=0.5411;
Vin=28;
Lm=170e-6;
Ts=1.11e-6;
Io=3;
n=3.0303;
Cc=3e-9;
Cr=64e-12;
P=[1e-6 2e-6 3e-6 4e-6 5e-6 6e-6 7e-6 8e-6 9e-6 10e-6] 
ilm=0.0989;

Vcc=(D/(1-D))*Vin;
Z=Vcc;
A=((Vin*D*Ts/Lm)+(Io/n))/Vcc;
B=Vin*Vin/(Vcc*Vcc*Lm*A);
C=Io/n;



fun=@(Lr)(((Vin*Cr*Lm)/(((((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)+((ilm*ilm*Lm*Cc)/(((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)))*(((1-D)*Ts)-((ilm*Lm)/(((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)))))+((Lr*(((((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)+((ilm*ilm*Lm*Cc)/(((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)))*(((1-D)*Ts)-((ilm*Lm)/(((Z*Cc)+((A+(B*Cr))*((B*Cr)+C)*Lr))/Cc)))))/(Lm*Vin))+((C*Lr)/Vin));

[Lr_opt,fval]=fminbnd(fun,0,10e-6,optimset('TolX',1e-10));


%% Plot the function and mark its minimum, as found by fminbnd

fplot(fun,[0.5,4]*1e-6);
hold on; plot(Lr_opt,fval,'rx','MarkerSize',10); hold off
ylim([0.8,1.7]*1e-7)

title("Variation with Lr")
xlabel("Lr")
ylabel("Interval II")

shg

[Prajakta Ghatage]

Thank you so much.

[Matt J]

You are quite welcome, but please Accept-click the answer to indicate that your question was addressed.