How to convert simulink based Reinforcement learning environment to function or script based.

Hi

I followed water tank model for reinforcement learning.

Since it is based on simulink , is there a way to implement this example in function or script based. I do not intend to use simulink.

Looking forward to hear from you.

Thanks

Chetan

Answer 1

[Sam Chak]

Hi @Chetan Thakur

I believe you can create a MATLAB function to describe the water level in a pump-driven storage tank, and then utilize it to construct the MATLAB Reinforcement Learning Environment.

tspan  = [0 20];
h0     =  0.2;      % initial height of water level
[t, h] = ode45(@watertank, tspan, h0);
plot(t, h), grid on
xlabel('Time'), ylabel('Water level')

% Water Tank System
function dhdt = watertank(t, h)
    A    = 1;   % cross-sectional area of the tank
    a    = 1;   % constant related to the flow rate out of the tank
    b    = 1;   % constant related to the flow rate into the tank
    V    = 1;   % voltage applied to the pump
    dhdt = - a/A*sqrt(h) + b/A*V;
end