CODEWITHSUNDEEP
matlabfor-loopsimulink
Ben
Ben

Reputation: 1837

Simulink model inside for loop

I want to check how some block parameter affect the results in my simulation.

I use a for loop to change this parameter however it works pretty slow.

mdl_name='Gains';
open_system(mdl_name);

for K1 = 1:2
    for K2 = 1:2
        for K3 = 0.1:0.1:2

            set_param([mdl_name,'/Tfn'],'Numerator',mat2str(50),'Denominator',mat2str([20]));
            set_param([mdl_name,'/K1'],'Gain',mat2str(K1));
            set_param([mdl_name,'/K2'],'Gain',mat2str(K2));
            set_param([mdl_name,'/K3'],'Gain',mat2str(K3));

            mdl_name='PWM_Gains';
            open_system(mdl_name);

            Data = sim(mdl_name,'StartTime','40','StopTime','85','FixedStep',num2str(dt));
            Delta_PWM{i} = Data.get('Delta_PWM').signals.values;

            Gains{i} = [K1,K2,K3];

            i = i + 1
        end
    end
end

Is there any more efficient way to solve it?

Thank you

Upvotes: 0

Views: 1316

Answers (1)

rinkert
rinkert

Reputation: 6863

Use parsim.

You can use load_system instead of open_system to avoid opening simulink. You probably only have to call it once, before the for loop, and not every iteration.

If it still takes too long, try larger stepsize (or variable, to be determined by the ode solver), or a larger step size in the changing parameters.

Example using parsim:

% load model
mdl_name='Gains';
load_system(mdl_name);

% parameter to simulate
K1 = 1:2;
K2 = 1:2;
K3 = 0.1:0.1:2;

% create matrix with all combinations
[k1, k2, k3] = ndgrid(K1, K2, K3);
Gains = [k1(:), k2(:), k3(:)];

% Create an array of SimulationInput objects and specify the sweep value for each simulation
numSims = size(Gains,1);
simIn(1:numSims) = Simulink.SimulationInput(model);

% setup models
for idx = 1:numSims
    % get Gains
    K1 = Gains(idx,1);
    K2 = Gains(idx,2);
    K3 = Gains(idx,3);

    % set Gains
    simIn(idx) = simIn(idx).setBlockParameter([mdl_name,'/Tfn'],'Numerator',mat2str(50),'Denominator',mat2str([20]));
    simIn(idx) = simIn(idx).setBlockParameter([mdl_name,'/K1'],'Gain',mat2str(K1));
    simIn(idx) = simIn(idx).setBlockParameter([mdl_name,'/K2'],'Gain',mat2str(K2));
    simIn(idx) = simIn(idx).setBlockParameter([mdl_name,'/K3'],'Gain',mat2str(K3));

    % set simulation parameters
    simIn(idx) = simIn(idx).setModelParameter('StartTime','40', 'StopTime','85','FixedStep',num2str(dt));
end

% Simulate the model 
simOut = parsim(simIn);

Upvotes: 1

Related Questions