setup of an initial condition using the do_mpc toolbox

Question

I am using the do_mpc toolbox in python. I want to set my own initial condition for the estimator (MHE) but the documentation is not very clear how I can do that. has anyone already done it? here the link of the toolbox and the specific class https://www.do-mpc.com/en/latest/api/do_mpc.estimator.MHE.opt_x_num.html#do_mpc.estimator.MHE.opt_x_num

Thank you very much for replying

score 0 · Answer 1 · answered May 14 '21 at 19:25

You have probably found an answer by now, but the set_initial_guess function of the estimator suggests that you can assign values to the opt_x_num variable directly:

def set_initial_guess(self):
    """..."""
    assert self.flags['setup'] == True, 'mhe was not setup yet. Please call mhe.setup().'

    self.opt_x_num['_x'] = self._x0.cat/self._x_scaling
    self.opt_x_num['_u'] = self._u0.cat/self._u_scaling
    self.opt_x_num['_z'] = self._z0.cat/self._z_scaling
    self.opt_x_num['_p_est'] = self._p_est0.cat/self._p_est_scaling

    self.flags['set_initial_guess'] = True

In particular, the above code samples sets each opt_x_num component to an instance of DM (from casadi). The values are computed using the x0, u0, etc. variables. As the docstrings suggest:

Uses the current class attributes :py:obj:x0, :py:obj:z0 and :py:obj:u0, :py:obj:p_est0 to create an initial guess for the MHE.

you can set the initial guess contents by setting the x0, u0, etc. variables. For example, you can call something like:

self._controller.x0 = np.array([100, 200, 0, 0])
self._controller.set_initial_guess()

Finally, you can ask for further help by opening a GitHub discussion - https://github.com/do-mpc/do-mpc/discussions

setup of an initial condition using the do_mpc toolbox

1 Answers1