qlibs::smithPredictor Class Reference

A Smith Predictor implementation for compensating time delays in control systems. More...

#include <ltisys.hpp>

Public Member Functions
virtual	~smithPredictor ()
	smithPredictor (ltisys &modelTf, ITransportDelay &mDelay, const real_t initialCondition=0.0_re)
	Constructs a Smith Predictor with a plant model, delay model, and optional initial output estimate.
	smithPredictor (ltisys &modelTf, ITransportDelay &mDelay, ltisys &filterTf, const real_t initialCondition=0.0_re)
	Constructs a Smith Predictor with a plant model, delay model, and optional initial output estimate.
bool	updatePrediction (const real_t ut, const real_t yt) noexcept
	Updates the internal prediction based on control input and measured plant output.
real_t	getPrediction () const noexcept
	Retrieves the current delay-free predicted output of the system.
bool	setFilter (ltisys &filterTf) noexcept
	Sets an optional filter for the internal Smith Predictor model.

Detailed Description

A Smith Predictor implementation for compensating time delays in control systems.

The Smith Predictor is a model-based feedforward control strategy designed to improve performance in systems with significant dead time. It estimates the delay-free output using an internal model of the process, a delay block, and optionally an output filter model.

Constructor & Destructor Documentation

~smithPredictor()

virtual qlibs::smithPredictor::~smithPredictor ( )

inlinevirtual

smithPredictor() [1/2]

qlibs::smithPredictor::smithPredictor ( ltisys & modelTf, ITransportDelay & mDelay, const real_t initialCondition = 0.0_re )

inline

Constructs a Smith Predictor with a plant model, delay model, and optional initial output estimate.

Parameters

[in]	modelTf	Reference to the LTI system model representing the delay-free plant.
[in]	mDelay	Reference to the transport delay block modeling the plant’s dead time.
[in]	initialCondition	Initial value for the internal output prediction (yp_hat). Default is 0.0.

Note

Both the model and delay block are passed by reference and stored internally as pointers.

smithPredictor() [2/2]

qlibs::smithPredictor::smithPredictor ( ltisys & modelTf, ITransportDelay & mDelay, ltisys & filterTf, const real_t initialCondition = 0.0_re )

inline

Constructs a Smith Predictor with a plant model, delay model, and optional initial output estimate.

Parameters

[in]	modelTf	Reference to the LTI system model representing the delay-free plant.
[in]	mDelay	Reference to the transport delay block modeling the plant’s dead time.
[in]	filterTf	Reference to the LTI system used as the robustness filter.
[in]	initialCondition	Initial value for the internal output prediction (yp_hat). Default is 0.0.

Note

Both the model and delay block are passed by reference and stored internally as pointers.

Member Function Documentation

getPrediction()

real_t qlibs::smithPredictor::getPrediction ( ) const

inlinenoexcept

Retrieves the current delay-free predicted output of the system.

Returns

The internally computed predicted output.

setFilter()

bool qlibs::smithPredictor::setFilter ( ltisys & filterTf )

inlinenoexcept

Sets an optional filter for the internal Smith Predictor model.

The filter is used to improve the system's robustness, attenuate measurement noise, and ensure internal stability, particularly in the case of unstable plant dynamics.

Parameters

[in]

filterTf

Reference to the LTI system used as the robustness filter.

Returns

true if the filter was set successfully.

Note

The filter is applied internally to the model-based prediction and does not affect the plant or delay block directly.

updatePrediction()

bool smithPredictor::updatePrediction ( const real_t ut, const real_t yt )

noexcept

Updates the internal prediction based on control input and measured plant output.

This method should be called at each control step. It updates the internal predicted output by propagating the input through the internal delay-free model and adjusting the prediction using the actual plant output.

Parameters

[in]	ut	The control input applied to the real plant.
[in]	yt	The actual measured output from the delayed plant.

Returns

true if the prediction was successfully updated, false otherwise.

Note

The time-step used to call this function must match the time base used by both the plant model and the delay block.