Well, first off let's do this with an individual parameter.
% Parameters
K = 1;
T = 1;
L = 1;
% Create the system
s = tf('s');
sys = (K/(T*s + 1))*exp(-L*s)
% Step response
step(sys)
You can instead place this in a big loop for different parameter values. For example:
% Parameters
K = [0.5,1.0,1.5];
T = [0.5,1.0,1.5];
L = [0.5,1.0,1.5];
% Initialize the loop counter
count = 1;
for ii = 1:numel(K)
for jj = 1:numel(T)
for kk = 1:numel(L)
% Create the system (use 3rd dimension to concatenate)
sys(:,:,count) = (K(ii)/(T(jj)*s + 1))*exp(-L(kk)*s);
count = count + 1;
end
end
end
% Step response
step(sys)
By the way, there is a way more elegant/efficient way of expressing parameter variations shown in this documentation link. However, it didn't seem to work because of the exponential portion, which can't accept a realp tunable parameter.
If you want to pursue the above, you may need to do a Pade approximation of that exponential time delay piece first.
- Sebastian
