Class KalmanEstimatorUtil
- java.lang.Object
-
- org.orekit.estimation.sequential.KalmanEstimatorUtil
-
public class KalmanEstimatorUtil extends Object
Utility class for Kalman Filter.This class includes common methods used by the different Kalman models in Orekit (i.e., Extended, Unscented, and Semi-analytical)
- Since:
- 11.3
-
-
Method Summary
All Methods Static Methods Concrete Methods Modifier and Type Method Description static <T extends ObservedMeasurement<T>>
voidapplyDynamicOutlierFilter(EstimatedMeasurement<T> measurement, RealMatrix innovationCovarianceMatrix)
Set and apply a dynamic outlier filter on a measurement.static void
checkDimension(int dimension, ParameterDriversList orbitalParameters, ParameterDriversList propagationParameters, ParameterDriversList measurementParameters)
Check dimension.static RealVector
computeInnovationVector(EstimatedMeasurement<?> predicted)
Compute the unnormalized innovation vector from the given predicted measurement.static RealVector
computeInnovationVector(EstimatedMeasurement<?> predicted, double[] sigma)
Compute the normalized innovation vector from the given predicted measurement.static MeasurementDecorator
decorate(ObservedMeasurement<?> observedMeasurement, AbsoluteDate referenceDate)
Decorate an observed measurement.static MeasurementDecorator
decorateUnscented(ObservedMeasurement<?> observedMeasurement, AbsoluteDate referenceDate)
Decorate an observed measurement for an Unscented Kalman Filter.static SpacecraftState[]
filterRelevant(ObservedMeasurement<?> observedMeasurement, SpacecraftState[] allStates)
Filter relevant states for a measurement.
-
-
-
Method Detail
-
decorate
public static MeasurementDecorator decorate(ObservedMeasurement<?> observedMeasurement, AbsoluteDate referenceDate)
Decorate an observed measurement.The "physical" measurement noise matrix is the covariance matrix of the measurement. Normalizing it consists in applying the following equation: Rn[i,j] = R[i,j]/σ[i]/σ[j] Thus the normalized measurement noise matrix is the matrix of the correlation coefficients between the different components of the measurement.
- Parameters:
observedMeasurement
- the measurementreferenceDate
- reference date- Returns:
- decorated measurement
-
decorateUnscented
public static MeasurementDecorator decorateUnscented(ObservedMeasurement<?> observedMeasurement, AbsoluteDate referenceDate)
Decorate an observed measurement for an Unscented Kalman Filter.This method uses directly the measurement's covariance matrix, without any normalization.
- Parameters:
observedMeasurement
- the measurementreferenceDate
- reference date- Returns:
- decorated measurement
- Since:
- 11.3.2
-
checkDimension
public static void checkDimension(int dimension, ParameterDriversList orbitalParameters, ParameterDriversList propagationParameters, ParameterDriversList measurementParameters)
Check dimension.- Parameters:
dimension
- dimension to checkorbitalParameters
- orbital parameterspropagationParameters
- propagation parametersmeasurementParameters
- measurements parameters
-
filterRelevant
public static SpacecraftState[] filterRelevant(ObservedMeasurement<?> observedMeasurement, SpacecraftState[] allStates)
Filter relevant states for a measurement.- Parameters:
observedMeasurement
- measurement to considerallStates
- all states- Returns:
- array containing only the states relevant to the measurement
-
applyDynamicOutlierFilter
public static <T extends ObservedMeasurement<T>> void applyDynamicOutlierFilter(EstimatedMeasurement<T> measurement, RealMatrix innovationCovarianceMatrix)
Set and apply a dynamic outlier filter on a measurement.Loop on the modifiers to see if a dynamic outlier filter needs to be applied.
Compute the sigma array using the matrix in input and set the filter.
Apply the filter by calling the modify method on the estimated measurement.
Reset the filter.
- Type Parameters:
T
- the type of measurement- Parameters:
measurement
- measurement to filterinnovationCovarianceMatrix
- So called innovation covariance matrix S, with:S = H.Ppred.Ht + R
Where:
- H is the normalized measurement matrix (Ht its transpose)
- Ppred is the normalized predicted covariance matrix
- R is the normalized measurement noise matrix
-
computeInnovationVector
public static RealVector computeInnovationVector(EstimatedMeasurement<?> predicted)
Compute the unnormalized innovation vector from the given predicted measurement.- Parameters:
predicted
- predicted measurement- Returns:
- the innovation vector
-
computeInnovationVector
public static RealVector computeInnovationVector(EstimatedMeasurement<?> predicted, double[] sigma)
Compute the normalized innovation vector from the given predicted measurement.- Parameters:
predicted
- predicted measurementsigma
- measurement standard deviation- Returns:
- the innovation vector
-
-