include/mpc.h File Reference

The user interface to the MPC algorithm and auxiliary functions. More...

#include "mpc_base.h"
#include "mpc_inc.h"
#include "mpc_stc.h"
#include "mpc_ref.h"

Go to the source code of this file.

Functions
void	mpc_ctl_solve_problem (struct mpc_ctl *ctl, const real_t x[])
void	mpc_predict_next_state (struct mpc_ctl *ctl, real_t x[])
void	mpc_ctl_form_qp (struct mpc_ctl *ctl, const real_t x[])

---

Detailed Description

The user interface to the MPC algorithm and auxiliary functions.

---

Function Documentation

void mpc_ctl_form_qp	(	struct mpc_ctl *	ctl,
		const real_t	x[]
	)

Form a quadratic program for the given system state vector.

Form a quadratic program qpx given a pointer to the MPC problem and the current state vector x. This function works in the same way as mpc_ctl_solve_problem, except that it only forms the QP to solve, without actually solving it. The resulting QP for the given state is stored in the structure ctl.qpx. The data stored in this structure can then be used together with other QP solvers. See struct mpc_qpx for the meaning of the fields of qpx.

Parameters:

ctl	contains the MPC problem information
x	the current state of the system

void mpc_ctl_solve_problem	(	struct mpc_ctl *	ctl,
		const real_t	x[]
	)

Compute the MPC optimal control sequence for the current state.

Compute the optimal control sequence u_opt given a pointer to the MPC controller ctl and the current state vector x. This function is a one-size-fits-all MPC solver. It automatically call the functions appropriate for the type of problem. Four cases are considered:

• input constraints, fixed setpoint
• input constraints, trajectory tracking
• state and input constraints, fixed setpoint
• state and input contraints, trajectory tracking

If the problem if either input only, or input and state constrained is determined at the time of automatically generating the code. The user cannot change this at runtime, and the algorithm being called is different for each case. Both of the reference tracking cases are specified at run-time, by setting both the ctl.x_ref and ctl.u_ref pointers to point to an array of appropriate size (HOR_STATES, and HOR_INPUTS, respectively). If any of them point to the zero pointer, the fixed setpoint cases are considered.

The computed optimal control sequence is saved in ctl.u_opt.

Parameters:

ctl	contains the MPC problem information
x	the current state of the system

void mpc_predict_next_state	(	struct mpc_ctl *	ctl,
		real_t	x[]
	)

Predict the state at the next sampling time.

Predict the state at the next sampling time given a pointer to the MPC controller ctl and the current state vector x. ctl contains the discrete-time system matrices Ad, and Bd and the optimal input sequence u_opt. Ad and Bd have been setup by the automatic code generation procedure, and u_opt is automatically computed by mpc_ctl_solve_problem function. x is an array with STATES elements. x will be overwritten with the predicted value according to x = Ad * x + Bd * u_opt. Note that only the first input vector of u_opt sequence is used.

Parameters:

ctl	contains the MPC problem information
x	the current state of the system