lsqnonlin() with Simulink

I am using lsqnonlin() with Simulink as it is used here. I am getting the errors:

1. Unable to perform assignment because the size of the left side is 30012-by-1 and the size of the right side is 30018-by-1
2. Error in finitedifferences
3. Error in computeFinDiffGradAndJac

Here is my code:

function [f] = trackq

% Computes f using lsqnonlin.

mdl = 'armDynamics';

open_system(mdl) % Load the model

in = Simulink.SimulationInput(mdl); % Create simulation input object

in = in.setModelParameter('StopTime','100'); % Stop time 100

f0 = 1; % Initial f

%

options = optimoptions(@lsqnonlin, 'Algorithm', 'levenberg-marquardt',...

'Display','off','StepTolerance',0.001,'OptimalityTolerance',0.001);

% Optimize f

set_param(mdl,'FastRestart','on'); % Fast restart

% f = lsqnonlin(@compute_error, f0, [], [], options);

f = lsqnonlin(@compute_error, f0, [], [], options);

set_param(mdl,'FastRestart','off');

function error = compute_error(f)

% Set the simulation input object parameters

in = in.setVariable('f', f, 'Workspace', mdl);

% Simulate

out = sim(in);

% get q from the actual space arm

armCoords = out.qArm.signals.values;

armCoords = armCoords(2:4);

armCoords = armCoords';

% get q from modelled space arm

qsim = out.get('yout');

qsim = qsim.getElement(1);

qsim = qsim.Values;

qsim = qsim.Data;

qsim = qsim';

error = armCoords - qsim;

end

end

Answer 1

[arushi]

Hi Kukhokuhle,

The error you’re encountering with lsqnonlin might be because there is a mismatch in the size of the arrays being subtracted in the compute_error function. Specifically, the armCoords array and the qsim array do not have the same number of elements when you attempt to compute the error by subtracting one from the other.

Here are a few ways you can try to resolve the issue:

1. Check Array Sizes: Ensure that the armCoords and qsim arrays have the same number of elements before performing the subtraction. You can add debugging statements to print their sizes just before the subtraction operation.
2. Consistent Output: The function you pass to lsqnonlin must return an array with a consistent number of elements.
3. Simulation Output: Verify that the simulation output out contains the expected data with the correct dimensions.
4. Finite Differences: The error in finitedifferences and computeFinDiffGradAndJac indicates that the Jacobian estimation step in lsqnonlin is failing due to inconsistent array sizes.Make sure that the function is vectorized properly and that it returns a vector of residuals.

Hope this helps.