KalmanEstimator.java
/* Copyright 2002-2019 CS Systèmes d'Information
* Licensed to CS Systèmes d'Information (CS) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* CS licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.orekit.estimation.sequential;
import java.util.List;
import org.hipparchus.exception.MathRuntimeException;
import org.hipparchus.filtering.kalman.ProcessEstimate;
import org.hipparchus.filtering.kalman.extended.ExtendedKalmanFilter;
import org.hipparchus.linear.MatrixDecomposer;
import org.hipparchus.linear.MatrixUtils;
import org.hipparchus.linear.RealMatrix;
import org.hipparchus.linear.RealVector;
import org.orekit.errors.OrekitException;
import org.orekit.estimation.measurements.ObservedMeasurement;
import org.orekit.estimation.measurements.PV;
import org.orekit.propagation.conversion.NumericalPropagatorBuilder;
import org.orekit.propagation.conversion.PropagatorBuilder;
import org.orekit.propagation.numerical.NumericalPropagator;
import org.orekit.time.AbsoluteDate;
import org.orekit.utils.ParameterDriver;
import org.orekit.utils.ParameterDriversList;
import org.orekit.utils.ParameterDriversList.DelegatingDriver;
/**
* Implementation of a Kalman filter to perform orbit determination.
* <p>
* The filter uses a {@link NumericalPropagatorBuilder} to initialize its reference trajectory {@link NumericalPropagator}.
* </p>
* <p>
* The estimated parameters are driven by {@link ParameterDriver} objects. They are of 3 different types:<ol>
* <li><b>Orbital parameters</b>:The position and velocity of the spacecraft, or, more generally, its orbit.<br>
* These parameters are retrieved from the reference trajectory propagator builder when the filter is initialized.</li>
* <li><b>Propagation parameters</b>: Some parameters modelling physical processes (SRP or drag coefficients etc...).<br>
* They are also retrieved from the propagator builder during the initialization phase.</li>
* <li><b>Measurements parameters</b>: Parameters related to measurements (station biases, positions etc...).<br>
* They are passed down to the filter in its constructor.</li>
* </ol>
* </p>
* <p>
* The total number of estimated parameters is m, the size of the state vector.
* </p>
* <p>
* The Kalman filter implementation used is provided by the underlying mathematical library Hipparchus.
* All the variables seen by Hipparchus (states, covariances, measurement matrices...) are normalized
* using a specific scale for each estimated parameters or standard deviation noise for each measurement components.
* </p>
*
* <p>A {@link KalmanEstimator} object is built using the {@link KalmanEstimatorBuilder#build() build}
* method of a {@link KalmanEstimatorBuilder}.</p>
*
* @author Romain Gerbaud
* @author Maxime Journot
* @author Luc Maisonobe
* @since 9.2
*/
public class KalmanEstimator {
/** Builders for numerical propagators. */
private List<NumericalPropagatorBuilder> propagatorBuilders;
/** Reference date. */
private final AbsoluteDate referenceDate;
/** Kalman filter process model. */
private final Model processModel;
/** Filter. */
private final ExtendedKalmanFilter<MeasurementDecorator> filter;
/** Observer to retrieve current estimation info. */
private KalmanObserver observer;
/** Kalman filter estimator constructor (package private).
* @param decomposer decomposer to use for the correction phase
* @param propagatorBuilders propagators builders used to evaluate the orbit.
* @param processNoiseMatricesProviders providers for process noise matrices
* @param estimatedMeasurementParameters measurement parameters to estimate
*/
KalmanEstimator(final MatrixDecomposer decomposer,
final List<NumericalPropagatorBuilder> propagatorBuilders,
final List<CovarianceMatrixProvider> processNoiseMatricesProviders,
final ParameterDriversList estimatedMeasurementParameters) {
this.propagatorBuilders = propagatorBuilders;
this.referenceDate = propagatorBuilders.get(0).getInitialOrbitDate();
this.observer = null;
// Build the process model and measurement model
this.processModel = new Model(propagatorBuilders, processNoiseMatricesProviders,
estimatedMeasurementParameters);
this.filter = new ExtendedKalmanFilter<>(decomposer, processModel, processModel.getEstimate());
}
/** Set the observer.
* @param observer the observer
*/
public void setObserver(final KalmanObserver observer) {
this.observer = observer;
}
/** Get the current measurement number.
* @return current measurement number
*/
public int getCurrentMeasurementNumber() {
return processModel.getCurrentMeasurementNumber();
}
/** Get the current date.
* @return current date
*/
public AbsoluteDate getCurrentDate() {
return processModel.getCurrentDate();
}
/** Get the "physical" estimated state (i.e. not normalized)
* @return the "physical" estimated state
*/
public RealVector getPhysicalEstimatedState() {
return processModel.getPhysicalEstimatedState();
}
/** Get the "physical" estimated covariance matrix (i.e. not normalized)
* @return the "physical" estimated covariance matrix
*/
public RealMatrix getPhysicalEstimatedCovarianceMatrix() {
return processModel.getPhysicalEstimatedCovarianceMatrix();
}
/** Get the orbital parameters supported by this estimator.
* <p>
* If there are more than one propagator builder, then the names
* of the drivers have an index marker in square brackets appended
* to them in order to distinguish the various orbits. So for example
* with one builder generating Keplerian orbits the names would be
* simply "a", "e", "i"... but if there are several builders the
* names would be "a[0]", "e[0]", "i[0]"..."a[1]", "e[1]", "i[1]"...
* </p>
* @param estimatedOnly if true, only estimated parameters are returned
* @return orbital parameters supported by this estimator
*/
public ParameterDriversList getOrbitalParametersDrivers(final boolean estimatedOnly) {
final ParameterDriversList estimated = new ParameterDriversList();
for (int i = 0; i < propagatorBuilders.size(); ++i) {
final String suffix = propagatorBuilders.size() > 1 ? "[" + i + "]" : null;
for (final ParameterDriver driver : propagatorBuilders.get(i).getOrbitalParametersDrivers().getDrivers()) {
if (driver.isSelected() || !estimatedOnly) {
if (suffix != null && !driver.getName().endsWith(suffix)) {
// we add suffix only conditionally because the method may already have been called
// and suffixes may have already been appended
driver.setName(driver.getName() + suffix);
}
estimated.add(driver);
}
}
}
return estimated;
}
/** Get the propagator parameters supported by this estimator.
* @param estimatedOnly if true, only estimated parameters are returned
* @return propagator parameters supported by this estimator
*/
public ParameterDriversList getPropagationParametersDrivers(final boolean estimatedOnly) {
final ParameterDriversList estimated = new ParameterDriversList();
for (PropagatorBuilder builder : propagatorBuilders) {
for (final DelegatingDriver delegating : builder.getPropagationParametersDrivers().getDrivers()) {
if (delegating.isSelected() || !estimatedOnly) {
for (final ParameterDriver driver : delegating.getRawDrivers()) {
estimated.add(driver);
}
}
}
}
return estimated;
}
/** Get the list of estimated measurements parameters.
* @return the list of estimated measurements parameters
*/
public ParameterDriversList getEstimatedMeasurementsParameters() {
return processModel.getEstimatedMeasurementsParameters();
}
/** Process a single measurement.
* <p>
* Update the filter with the new measurement by calling the estimate method.
* </p>
* @param observedMeasurement the measurement to process
* @return estimated propagators
*/
public NumericalPropagator[] estimationStep(final ObservedMeasurement<?> observedMeasurement) {
try {
final ProcessEstimate estimate = filter.estimationStep(decorate(observedMeasurement));
processModel.finalizeEstimation(observedMeasurement, estimate);
if (observer != null) {
observer.evaluationPerformed(processModel);
}
return processModel.getEstimatedPropagators();
} catch (MathRuntimeException mrte) {
throw new OrekitException(mrte);
}
}
/** Process several measurements.
* @param observedMeasurements the measurements to process in <em>chronologically sorted</em> order
* @return estimated propagators
*/
public NumericalPropagator[] processMeasurements(final Iterable<ObservedMeasurement<?>> observedMeasurements) {
NumericalPropagator[] propagators = null;
for (ObservedMeasurement<?> observedMeasurement : observedMeasurements) {
propagators = estimationStep(observedMeasurement);
}
return propagators;
}
/** Decorate an observed measurement.
* <p>
* 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.
* </p>
* @param observedMeasurement the measurement
* @return decorated measurement
*/
private MeasurementDecorator decorate(final ObservedMeasurement<?> observedMeasurement) {
// Normalized measurement noise matrix contains 1 on its diagonal and correlation coefficients
// of the measurement on its non-diagonal elements.
// Indeed, the "physical" measurement noise matrix is the covariance matrix of the measurement
// Normalizing it leaves us with the matrix of the correlation coefficients
final RealMatrix covariance;
if (observedMeasurement instanceof PV) {
// For PV measurements we do have a covariance matrix and thus a correlation coefficients matrix
final PV pv = (PV) observedMeasurement;
covariance = MatrixUtils.createRealMatrix(pv.getCorrelationCoefficientsMatrix());
} else {
// For other measurements we do not have a covariance matrix.
// Thus the correlation coefficients matrix is an identity matrix.
covariance = MatrixUtils.createRealIdentityMatrix(observedMeasurement.getDimension());
}
return new MeasurementDecorator(observedMeasurement, covariance, referenceDate);
}
}