bdm::LQG Class Reference

Detailed Description

Linear Quadratic Gaussian designer for constant penalizations and constant target Its internals are very close to Kalman estimator

#include <ctrlbase.h>

Public Member Functions
void	set_system (shared_ptr< StateSpace< fsqmat > > S0)
	set system parameters from given matrices
void	set_control_parameters (const mat &Qy0, const mat &Qu0, const vec &y_req0, int horizon0)
	set penalization matrices and control horizon
void	initialize ()
	set system parameters from Kalman filter
void	validate ()
virtual void	update_state ()
	function for future use which is called at each time td; Should call initialize()!
void	ricatti_step ()
	redesign one step of the
void	redesign ()
	Redesign control strategy.
vec	apply (const vec &state, const vec &ukm)
virtual vec	apply (const vec &cond)
	apply control strategy to obtain control input
Protected Attributes
shared_ptr< StateSpace< fsqmat > >	S
	StateSpace model from which we read data.
vec	y_req
	required value of the output y at time t (assumed constant)
vec	u_req
	required value of the output y at time t (assumed constant)
int	horizon
	Control horizon, set to maxint for infinite horizons.
mat	Qy
	penalization matrix Qy
mat	Qu
	penalization matrix Qu
int	td
	time of the design step - from horizon->0
mat	L
	controller parameters
temporary storage for ricatti - use initialize
convenience parameters
int	dimx
int	dimy
	convenience parameters
int	dimu
	convenience parameters
mat	pr
	parameters
mat	qux
	penalization
mat	qyx
	penalization
mat	s
	internal quadratic form
mat	qy
	penalization
mat	hqy
	pre_qr part
mat	pre_qr
	pre qr matrix
mat	post_qr
	post qr matrix

void bdm::LQG::initialize ( )

set system parameters from Kalman filter

refresh temporary storage - inefficient can be improved

The documentation for this class was generated from the following files: