Create the system module¶

The sytem module can also be created in MATLAB. Simply define the required matrices (e.g. Ac, Bc, Q, u_lb, e_lb) and save them in a mat file. In this example, we will assume the sys_motor.mat system module was created, based on the MPC setup described on A basic MPC problem.

Note: to read mat files, SciPy is required.

Generate the code¶

At first you have to generate the code that represents the MATLAB interface. Simply add the option matlab=True to the code generation function. The following Python commands will generate the MATLAB/Simulink interfaces for the sys_motor.mat example:

import muaompc
mpc = muaompc.ltidt.setup_mpc_problem('sys_motor.mat')
mpc.generate_c_files(matlab=True)

The generated cmpc directory now contains two new directories called matlab and simulink besides the usual C files. It contains all the code of the MATLAB and Simulink interfaces. A detailed description follows.

Compiling the interface¶

Once the code the interfaces have been created, the next step is to compile the MATLAB interface. Start MATLAB, set the generated cmpc/matlab directory as the working directory and call make, which will execute the make.m script. For example, assuming you are in the directory where you generated the data, inside MATLAB type:

>> cd cmpc/matlab
>> make

After this the compiled code is put inside the @mpc_ctl directory.

Adding the interface to MATLAB’s path¶

The last step is to add the matlab directory to the PATH environment in MATLAB.

Using the interface¶

Now you can use the interface which is encapsulated in a class called mpc_ctl, which represents the MPC controller. Simply declare an instance of that class, which we usually call ctl (controller). For example:

ctl = mpc_ctl;  % create an instance of the class
ctl.conf.in_iter = 10;  % configure the algorithm
x = [0.1; -0.5];  % current state
ctl.form_qp(x)  % form the QP for the current state
qpx = ctl.qpx;  % qpx contains the QP in standard form
u = quadprog(qpx.HoL, qpx.gxoL, [], [], [], [], qpx.u_lb, qpx.u_ub);
ctl.solve_problem(x);  % solve the MPC problem for current state
ctl.u_opt  % display the computed input sequence
norm(u - ctl.u_opt)  % ctl.u_opt is the MPC approximation of u

Example¶

The complete example can be found under the examples/ltidt/matlab_interface folder.

Compiling the interface¶

As explained above, to generate preconfigured Simulink model you need to pass matlab=True parameter to the generator call. The model is placed in cmpc/simulink folder. To run the model follow the next steps:

1. In MATLAB change the current folder to cmpc/simulink.
2. Open cmpc/simulink/simulinkmpc.mdl in Simulink.
3. Double click the S-Function Builder block to open the block properties window.
4. Click Build to build the S-Function.

Now the model is ready to be extended and used in your design.

Example¶

The example system simulink_example.mdl is located in the folder examples/ltidt/matlab_interface. It shows a sample use of a MPC controller. To run the example follow the next steps:

1. Set the current MATLAB folder to examples/ltidt/matlab_interface.
2. Type open_simulink_example in the MATLAB console.
3. Press the Build button in the block properties window which will appear on the screen.
4. Run the simulation.

The example includes the functionality of solving the control problem and prediction of the next state of the plant. The model can operate using closed-loop and open-loop mode. Also the model shows how to check the execution speed on the device.