Plant system identification via DE. Port of ControlPathParamEstimation.ts. More...

#include <control_path_param_estimation.hpp>

Inheritance diagram for libstp::autotune::control::ControlPathParamEstimation:

Collaboration diagram for libstp::autotune::control::ControlPathParamEstimation:

Public Types
using	MeasPoint = std::pair<double, double>

Public Member Functions
	ControlPathParamEstimation (ControlPath &controlPath, const std::vector< MeasPoint > &measData)
std::vector< double >	optimize (double limitMinParams, double limitMaxParams, bool dither, int maxNumSteps, double stepHeight, double stepTime, std::uint32_t seed=0xC0FFEEu)
	Run DE to fit the control path to the measured data.
double	rateCandidate (const std::vector< double > &paramList) override
	DE cost callback: squared error of the simulated vs. measured response.
double	dt () const
	Mean sample time derived from the measurement timestamps.
void	setStep (double stepHeight, double stepTime)
Public Member Functions inherited from libstp::autotune::control::IDifferentialEvolutionRating
virtual	~IDifferentialEvolutionRating ()=default

Detailed Description

Plant system identification via DE. Port of ControlPathParamEstimation.ts.

Given measured step-response data (time, value) pairs, finds the control-path parameters that minimise the squared error between the simulated and measured output. The sample time dt is derived as the mean spacing of the measurement timestamps, exactly as in the TypeScript.

Implements IDifferentialEvolutionRating so the DE optimizer can rate candidate parameter vectors.

Definition at line 24 of file control_path_param_estimation.hpp.

Member Typedef Documentation

◆ MeasPoint

using libstp::autotune::control::ControlPathParamEstimation::MeasPoint = std::pair<double, double>

Definition at line 27 of file control_path_param_estimation.hpp.

Constructor & Destructor Documentation

◆ ControlPathParamEstimation()

libstp::autotune::control::ControlPathParamEstimation::ControlPathParamEstimation	(	ControlPath &	controlPath,
		const std::vector< MeasPoint > &	measData )

Member Function Documentation

◆ dt()

double libstp::autotune::control::ControlPathParamEstimation::dt ( ) const

inlinenodiscard

Mean sample time derived from the measurement timestamps.

Definition at line 48 of file control_path_param_estimation.hpp.

◆ optimize()

std::vector< double > libstp::autotune::control::ControlPathParamEstimation::optimize	(	double	limitMinParams,
		double	limitMaxParams,
		bool	dither,
		int	maxNumSteps,
		double	stepHeight,
		double	stepTime,
		std::uint32_t	seed = 0xC0FFEEu )

Run DE to fit the control path to the measured data.

Returns: best-fit parameter vector.

◆ rateCandidate()

double libstp::autotune::control::ControlPathParamEstimation::rateCandidate ( const std::vector< double > & paramList )

overridevirtual

DE cost callback: squared error of the simulated vs. measured response.

Implements libstp::autotune::control::IDifferentialEvolutionRating.

◆ setStep()

void libstp::autotune::control::ControlPathParamEstimation::setStep	(	double	stepHeight,
		double	stepTime )

inline

Set the step parameters that rateCandidate() uses (normally set by optimize()). Exposed for direct cost-function testing.

Definition at line 52 of file control_path_param_estimation.hpp.

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

/home/runner/work/raccoon-lib/raccoon-lib/modules/libstp-autotune/include/autotune/control/control_path_param_estimation.hpp

Public Types

Public Member Functions