SmpcController Class Reference

Collaboration diagram for SmpcController:

Public Member Functions
	SmpcController (Forecaster *myForecaster, Engine *myEngine, SmpcConfiguration *mySmpcConfig)
	SmpcController (string pathToConfigFile)
void	initialiseSmpcController ()
void	controllerSmpc ()
uint_t	controlAction (real_t *u)
uint_t	controlAction (fstream &controlOutputJson)
DwnNetwork *	getDwnNetwork ()
ScenarioTree *	getScenarioTree ()
SmpcConfiguration *	getSmpcConfiguration ()
Forecaster *	getForecaster ()
Engine *	getEngine ()
void	moveForewardInTime ()
real_t	getEconomicKpi (uint_t simulationTime)
real_t	getSmoothKpi (uint_t simulationTime)
real_t	getNetworkKpi (uint_t simulationTime)
real_t	getSafetyKpi (uint_t simulationTime)
void	updateKpi (real_t *state, real_t *control)
	~SmpcController ()

Protected Member Functions
void	dualExtrapolationStep (real_t lambda)
void	solveStep ()
void	proximalFunG ()
void	dualUpdate ()
uint_t	algorithmApg ()
void	updateSmpcConfiguration (real_t *updateState, real_t *control, real_t *demand)

Protected Attributes
Engine *	ptrMyEngine
Forecaster *	ptrMyForecaster
SmpcConfiguration *	ptrMySmpcConfig
real_t *	devVecX
real_t *	devVecU
real_t *	devVecV
real_t *	devVecXi
real_t *	devVecPsi
real_t *	devVecAcceleratedXi
real_t *	devVecAcceleratedPsi
real_t *	devVecPrimalXi
real_t *	devVecPrimalPsi
real_t *	devVecDualXi
real_t *	devVecDualPsi
real_t *	devVecUpdateXi
real_t *	devVecUpdatePsi
real_t **	devPtrVecX
real_t **	devPtrVecU
real_t **	devPtrVecV
real_t **	devPtrVecAcceleratedPsi
real_t **	devPtrVecAcceleratedXi
real_t **	devPtrVecPrimalPsi
real_t **	devPtrVecPrimalXi
real_t *	devVecPrimalInfeasibilty
real_t *	devVecQ
real_t *	devVecR
real_t **	devPtrVecQ
real_t **	devPtrVecR
real_t *	devControlAction
real_t *	devStateUpdate
real_t	stepSize
bool	factorStepFlag
bool	simulatorFlag
real_t *	vecPrimalInfs
real_t	economicKpi
real_t	smoothKpi
real_t	safeKpi
real_t	networkKpi

Detailed Description

GPU-based stochastic model predictive controller for management of the drinking water water network. This method is used dual proximal gradient method to solve the optimisation problem resulted from the stochastic mpc controller. This algorithm is completely parallelisable takes full advantage of parallel computational capability of the GPU devices.

The object initiates the controller object. This contains

• The forecaster object to predict the nominal water demand and nominal prices
• The Engine object generates matrices that are used in the actual algorithm
• The SmpcConfiguration object that have all the user defined parameters of the algorithm

Constructor & Destructor Documentation

SmpcController::SmpcController	(	Forecaster *	myForecaster,
		Engine *	myEngine,
		SmpcConfiguration *	mySmpcConfig
	)

Construct a new Controller with a given engine.

Parameters

myForecaster	An instance of the Forecaster object
myEngine	An instance of Engine object
mySmpcConfig	An instance of the Smpc controller configuration object

Here is the call graph for this function:

SmpcController::SmpcController

(

string

pathToConfigFile

)

Construct a new Controller with a given engine.

Parameters

pathToConfigFile

path to the controller configuration file

Here is the call graph for this function:

SmpcController::~SmpcController

(

)

Destructor. Frees allocated memory.

Member Function Documentation

uint_t SmpcController::algorithmApg ( )

protected

This method executes the APG algorithm and returns the primal infeasibility.

Returns

primalInfeasibilty;

Here is the call graph for this function:

Here is the caller graph for this function:

uint_t SmpcController::controlAction

(

real_t *

u

)

Computes a control action and returns a status code which is an integer (1 = success).

Parameters

u	pointer to computed control action (CPU variable)

Returns

status code

Here is the call graph for this function:

Here is the caller graph for this function:

uint_t SmpcController::controlAction

(

fstream &

controlOutputJson

)

Compute the control action, stores in the json file provided to it and returns a status code (1 = success).

Parameters

controlJson

file pointer to the output json file

Returns

status code

Here is the call graph for this function:

void SmpcController::controllerSmpc

(

)

Invoke the SMPC controller on the current state of the network. This method invokes #updateStateControl, eliminateInputDistubanceCoupling and finally algorithmApg.

Here is the call graph for this function:

void SmpcController::dualExtrapolationStep ( real_t  lambda )

protected

Performs the dual extrapolation step with given parameter.

Parameters

extrapolation

parameter.

Here is the call graph for this function:

Here is the caller graph for this function:

void SmpcController::dualUpdate ( )

protected

Performs the update of the dual vector.

Here is the call graph for this function:

Here is the caller graph for this function:

DwnNetwork * SmpcController::getDwnNetwork

(

)

Get's the network object

Returns

DwnNetwork

Here is the call graph for this function:

Here is the caller graph for this function:

real_t SmpcController::getEconomicKpi

(

uint_t

simulationTime

)

Here is the caller graph for this function:

Engine * SmpcController::getEngine

(

)

Get's the Engine object

Returns

Engine

Forecaster * SmpcController::getForecaster

(

)

Get's the Forecaster object

Returns

Forecaster

Get's the forecaster object

Returns

Forecaster

Here is the caller graph for this function:

real_t SmpcController::getNetworkKpi

(

uint_t

simulationTime

)

Here is the call graph for this function:

Here is the caller graph for this function:

real_t SmpcController::getSafetyKpi

(

uint_t

simulationTime

)

Here is the caller graph for this function:

ScenarioTree * SmpcController::getScenarioTree

(

)

Get's the scenario tree object

Returns

scenarioTree

Here is the call graph for this function:

real_t SmpcController::getSmoothKpi

(

uint_t

simulationTime

)

Here is the caller graph for this function:

SmpcConfiguration * SmpcController::getSmpcConfiguration

(

)

Get's the Smpc controller configuration object

Returns

SmpcConfiguration

void SmpcController::initialiseSmpcController

(

)

Performs the initialise the smpc controller

• update the current state and previous controls in the device memory
• perform the factor step

Here is the call graph for this function:

Here is the caller graph for this function:

void SmpcController::moveForewardInTime

(

)

Here is the call graph for this function:

Here is the caller graph for this function:

void SmpcController::proximalFunG ( )

protected

Computes the proximal operator of g at the current point and updates (primal psi, primal xi) - Hx, (dual psi, dual xi) - z.

Here is the call graph for this function:

Here is the caller graph for this function:

void SmpcController::solveStep ( )

protected

Computes the dual gradient.This is the main computational algorithm for the proximal gradient algorithm

Here is the call graph for this function:

Here is the caller graph for this function:

void SmpcController::updateKpi	(	real_t *	state,
		real_t *	control
	)

update the KPI at the current time instance

Here is the call graph for this function:

Here is the caller graph for this function:

void SmpcController::updateSmpcConfiguration ( real_t *  updateState, real_t *  control, real_t *  demand  )

protected

When the SIMULATOR FLAG is set to 1, the previousControl, currentState and previousDemand vectors in the smpc controller configuration file are set.

Parameters

updateState	update the currentX in the controller configuration json
control	update the prevU in the controller configuration json
demand	update the prevDemand in the controller configuration json

Update tje json file using the commands from rapidJson functions When the SIMULATOR FLAG is set to 1, the previousControl, currentState and previousDemand vectors in the smpc controller configuration file are set.

Here is the call graph for this function:

Member Data Documentation

real_t* SmpcController::devControlAction

protected

real_t** SmpcController::devPtrVecAcceleratedPsi

protected

Pointer array for device pointers of accelerated psi

real_t** SmpcController::devPtrVecAcceleratedXi

protected

Pointer array for device pointers of accelerated xi

real_t** SmpcController::devPtrVecPrimalPsi

protected

Pointer array for device pointers of primal psi

real_t** SmpcController::devPtrVecPrimalXi

protected

Pointer array for device pointer of primal xi

real_t** SmpcController::devPtrVecQ

protected

Pointer array for device pointers of cost Q

real_t** SmpcController::devPtrVecR

protected

Pointer array for device pointers of cost R

real_t** SmpcController::devPtrVecU

protected

Pointer array for device pointers of control

real_t** SmpcController::devPtrVecV

protected

Pointer array for device pointers of reduced decision variable

real_t** SmpcController::devPtrVecX

protected

Pointer array for device pointers of state

real_t* SmpcController::devStateUpdate

protected

Updated state for the simulator

real_t* SmpcController::devVecAcceleratedPsi

protected

Pointer to device accelerated Psi

real_t* SmpcController::devVecAcceleratedXi

protected

Pointer to device accelerated Xi

real_t* SmpcController::devVecDualPsi

protected

Pointer to device dual Psi

real_t* SmpcController::devVecDualXi

protected

Pointer to device dual Xi

real_t* SmpcController::devVecPrimalInfeasibilty

protected

Pointer array for primal infeasibility Hx-z

real_t* SmpcController::devVecPrimalPsi

protected

Pointer to device primal Psi

real_t* SmpcController::devVecPrimalXi

protected

Pointer to device primal Xi

real_t* SmpcController::devVecPsi

protected

Pointer to device Psi

real_t* SmpcController::devVecQ

protected

Pointer for cost Q

real_t* SmpcController::devVecR

protected

Pointer for cost R

real_t* SmpcController::devVecU

protected

Pointer to device control

real_t* SmpcController::devVecUpdatePsi

protected

Pointer to device update psi

real_t* SmpcController::devVecUpdateXi

protected

Pointer to device updated xi

real_t* SmpcController::devVecV

protected

Pointer to device reduced control

real_t* SmpcController::devVecX

protected

Pointer to device state

real_t* SmpcController::devVecXi

protected

Pointer to device Xi

real_t SmpcController::economicKpi

protected

bool SmpcController::factorStepFlag

protected

Flag Factor step

real_t SmpcController::networkKpi

protected

Engine* SmpcController::ptrMyEngine

protected

Pointer to an Engine object. The Engine is responsible for the factor step.

Forecaster* SmpcController::ptrMyForecaster

protected

Pointer to the Forecaster object This contains the forecast of the demand and prices

SmpcConfiguration* SmpcController::ptrMySmpcConfig

protected

Pointer to the Smpc configuration object This contains the configurations for the controller

real_t SmpcController::safeKpi

protected

bool SmpcController::simulatorFlag

protected

real_t SmpcController::smoothKpi

protected

real_t SmpcController::stepSize

protected

Step size

real_t* SmpcController::vecPrimalInfs

protected

---

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

• SmpcController.cuh
• SmpcController.cu