Vectors must be the same length.

Hi,

I am getting this error here:

Error using plot

Vectors must be the same length.

Error in test (line 32)

plot(t,u);ylabel('Input'); axis([-202 max(t) [-201 202]*max(u)*1.1])

The code I am using is the below:

dt = 0.01 % sample rate

sim('speed')       % run the simulation it will save the data as y and u - 


dat =iddata(y,u,dt);    % IDDATA(output,input,sample_time)
order = [2 1 1] % ["number of a" "number of b" "pure delay"]

mod = arx(dat,order)    % Mod = ARX(DATA,ORDER)

a = mod.a; b=mod.b;     % a is the denominatar, b the numerator polynomial in z-1

yhat = filter(b,a,u); % calculate the filter output (i.e. y = b/a * u)
e = y-yhat;           % calculate the error


close all; % close all old figure windows
scrsz = get(groot,'ScreenSize');    % How big is the screen (so we can come up with a figure window size)
fig=figure('Position',[50 100 scrsz(3)*2/5 scrsz(4)*3/4]);

% Plot the input
subplot('position',[0.1 0.8 0.85 0.15]);
plot(t,u);ylabel('Input'); axis([-202 max(t) [-201 202]*max(u)*1.1])
name=['System ID Results for Model Order = ' num2str(order)];
title(name);
% Plot the output and the model
subplot('position',[0.1 0.35 0.85 0.40]);
plot(t,y,t,yhat); 
ylabel('Output');legend('Actual - y','Model - yhat','Location','NW');
% Plot the error
subplot('position',[0.1 0.1 0.85 0.20]); 
plot(t,e);ylabel('Model Error');xlabel('time [s]'); 

% some methods of assessing the model:
Sum_of_Squared_Errors = e'*e                                           % the model with smallest number is "best" fit
Coefficient_of_determination = 1-(e'*e)/((y-mean(y))'*(y-mean(y)))     % (range 0 to 1) will tend to 1 as the model fit improves
AIC = aic(mod)                                                         % look up in help - trades off fit to the data against number of parameters

figure(2);histfit(e,21,'normal') % plot the error distribution 
title('Distribution of Model Error');
xlabel('Error')

Thanks in advance!

[Torsten]

Check

size(t)
size(u)

before the command

plot(t,u)

If they are not equal, an error will pop up.

[I Bos]

Ah ok, I see. You are correct the size is not equal:

ans = 201 1

33 size(t)

ans = 501 1

How can use the same size to resolve this isue?

[Image Analyst]

You need to get to the root of the problem. WHY are they not the same size even though you expect them to be?

[I Bos]

thanks for your help, I think it is expected to not be the same size as the data keep changing.

I have used the numel function to resolve this matter and it seems that it works

Is there any alternative solution apart from numel?

[AndresVar]

@I Bos numel (or length) is the typical way.

In your case, the time vector spans 5 seconds but your data is less than 5 seconds.

So either you defined time vector arbitrarily or data is missing?

BUT if you are confident that the sample rate is 1/dt, then you can just redifine a time vector.

u=rand(10,1);

t0=0; % initial time
dt=0.01; 
fs=1/dt;

t2 = t0+(0:numel(u)-1)/fs;

An alternative is to pad data with 0's if you know data is missing at the front or at the end: Pad array - MATLAB padarray (mathworks.com)

[I Bos]

@AndresVar thanks very much for your help.

How the Padding works in my case, as I would need to add 300 '0's at the end of the u.

Many thanks again.

[Walter Roberson]

You would use a "direction" of 'post' to add at the end of something.