NMPC for a simple tank

4 views (last 30 days)
Nivedhitha S
Nivedhitha S on 26 Jul 2021
I tried to implement a NMPC for a simple tank system. Reference for the output variable (tank height) was say: 50 m, no matter how much I tried different combinations of prediction and control horizons and sample time, I couldn't achieve desired output (reference tracking).
Sample code:
% Init
nlobj_Tank = nlmpc(1,1,1);
nlobj_Tank.Model.StateFcn = @StateFunc_Tank;
Ts = 0.02;
nlobj_4Tank.Ts = Ts;
nlobj_4Tank.ControlHorizon = 2;
nlobj_4Tank.PredictionHorizon = 10;
% Validate
x0 = 10;
mv0 = 100;
validateFcns(nlobj_Tank, x0, mv0)
% Bounds
nlobj_Tank.ManipulatedVariables(1).Min = -1;
nlobj_Tank.ManipulatedVariables(1).Max = 800;
nlobj_Tank.ManipulatedVariables(1).RateMin = -0.1;
nlobj_Tank.ManipulatedVariables(1).RateMax = 100;
nlobj_4Tank.States(1).Min = 0;
nlobj_4Tank.States(1).Max = 1000;
% Actual plant model
Plant_state = @(x, u) StateFunc_4Tank(x, u) + 0.01*rand(1,1);
% Setpoint or Output reference
Out_ref = [50];
all_MV = zeros(25,1);
all_X_plant = zeros(25,1);
mv_now = mv0;
x_now = x0;
for i = 2:1:25
[mv_now, opt, info] = nlmpcmove(nlobj_Tank, x_now, mv_now, Out_ref);
all_MV(i,:) = mv_now;
x_now = Plant_state(x_now, mv_now);
all_X_plant(i,:) = x_now;
end
Code for tank: StateFunc_Tank;
%% Trying the single tank NMPC
z= zeros(1,1);
a=10; %Area of the tank
R=0.1; %resistance of valve
%dh/dt = Fin - R sqrt(h)
z(1) = (1/a)*(u(1)-(R*sqrt(x(1))));
I am quite new for control and MPC implementation. Can somebody guide me how to implement a NMPC? What are the tuning parameters/ methodology to do the same? How to validate and visualize the NMPC results?
For MPC we have MPC desginer app and many tutorials but for NMPC I couldn't find any resources. Any help would be great.
Thanks

Answers (0)

Categories

Find more on Model Predictive Control Toolbox in Help Center and File Exchange

Products


Release

R2021a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!