src/daComposant/daAlgorithms/KalmanFilter.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # Copyright (C) 2008-2019 EDF R&D
   4 #
   5 # This library is free software; you can redistribute it and/or
   6 # modify it under the terms of the GNU Lesser General Public
   7 # License as published by the Free Software Foundation; either
   8 # version 2.1 of the License.
   9 #
  10 # This library is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13 # Lesser General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU Lesser General Public
  16 # License along with this library; if not, write to the Free Software
  17 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
  18 #
  19 # See http://www.salome-platform.org/ or email : webmaster.salome@opencascade.com
  20 #
  21 # Author: Jean-Philippe Argaud, jean-philippe.argaud@edf.fr, EDF R&D
  22
  23 import logging
  24 from daCore import BasicObjects
  25 import numpy
  26
  27 # ==============================================================================
  28 class ElementaryAlgorithm(BasicObjects.Algorithm):
  29     def __init__(self):
  30         BasicObjects.Algorithm.__init__(self, "KALMANFILTER")
  31         self.defineRequiredParameter(
  32             name     = "EstimationOf",
  33             default  = "State",
  34             typecast = str,
  35             message  = "Estimation d'etat ou de parametres",
  36             listval  = ["State", "Parameters"],
  37             )
  38         self.defineRequiredParameter(
  39             name     = "StoreInternalVariables",
  40             default  = False,
  41             typecast = bool,
  42             message  = "Stockage des variables internes ou intermédiaires du calcul",
  43             )
  44         self.defineRequiredParameter(
  45             name     = "StoreSupplementaryCalculations",
  46             default  = [],
  47             typecast = tuple,
  48             message  = "Liste de calculs supplémentaires à stocker et/ou effectuer",
  49             listval  = [
  50                 "APosterioriCorrelations",
  51                 "APosterioriCovariance",
  52                 "APosterioriStandardDeviations",
  53                 "APosterioriVariances",
  54                 "BMA",
  55                 "CurrentState",
  56                 "CostFunctionJ",
  57                 "CostFunctionJb",
  58                 "CostFunctionJo",
  59                 "Innovation",
  60                 "PredictedState",
  61                 ]
  62             )
  63         self.requireInputArguments(
  64             mandatory= ("Xb", "Y", "HO", "R", "B" ),
  65             optional = ("U", "EM", "CM", "Q"),
  66             )
  67
  68     def run(self, Xb=None, Y=None, U=None, HO=None, EM=None, CM=None, R=None, B=None, Q=None, Parameters=None):
  69         self._pre_run(Parameters, Xb, Y, R, B, Q)
  70         #
  71         if self._parameters["EstimationOf"] == "Parameters":
  72             self._parameters["StoreInternalVariables"] = True
  73         #
  74         # Opérateurs
  75         # ----------
  76         Ht = HO["Tangent"].asMatrix(Xb)
  77         Ha = HO["Adjoint"].asMatrix(Xb)
  78         #
  79         if self._parameters["EstimationOf"] == "State":
  80             Mt = EM["Tangent"].asMatrix(Xb)
  81             Ma = EM["Adjoint"].asMatrix(Xb)
  82         #
  83         if CM is not None and "Tangent" in CM and U is not None:
  84             Cm = CM["Tangent"].asMatrix(Xb)
  85         else:
  86             Cm = None
  87         #
  88         # Nombre de pas identique au nombre de pas d'observations
  89         # -------------------------------------------------------
  90         if hasattr(Y,"stepnumber"):
  91             duration = Y.stepnumber()
  92         else:
  93             duration = 2
  94         #
  95         # Précalcul des inversions de B et R
  96         # ----------------------------------
  97         if self._parameters["StoreInternalVariables"] \
  98             or self._toStore("CostFunctionJ") \
  99             or self._toStore("CostFunctionJb") \
 100             or self._toStore("CostFunctionJo"):
 101             BI = B.getI()
 102             RI = R.getI()
 103         #
 104         # Initialisation
 105         # --------------
 106         Xn = Xb
 107         Pn = B
 108         #
 109         self.StoredVariables["Analysis"].store( Xn.A1 )
 110         if self._toStore("APosterioriCovariance"):
 111             self.StoredVariables["APosterioriCovariance"].store( Pn.asfullmatrix(Xn.size) )
 112             covarianceXa = Pn
 113         Xa               = Xn
 114         previousJMinimum = numpy.finfo(float).max
 115         #
 116         for step in range(duration-1):
 117             if hasattr(Y,"store"):
 118                 Ynpu = numpy.asmatrix(numpy.ravel( Y[step+1] )).T
 119             else:
 120                 Ynpu = numpy.asmatrix(numpy.ravel( Y )).T
 121             #
 122             if U is not None:
 123                 if hasattr(U,"store") and len(U)>1:
 124                     Un = numpy.asmatrix(numpy.ravel( U[step] )).T
 125                 elif hasattr(U,"store") and len(U)==1:
 126                     Un = numpy.asmatrix(numpy.ravel( U[0] )).T
 127                 else:
 128                     Un = numpy.asmatrix(numpy.ravel( U )).T
 129             else:
 130                 Un = None
 131             #
 132             if self._parameters["EstimationOf"] == "State":
 133                 Xn_predicted = Mt * Xn
 134                 if Cm is not None and Un is not None: # Attention : si Cm est aussi dans M, doublon !
 135                     Cm = Cm.reshape(Xn.size,Un.size) # ADAO & check shape
 136                     Xn_predicted = Xn_predicted + Cm * Un
 137                 Pn_predicted = Q + Mt * Pn * Ma
 138             elif self._parameters["EstimationOf"] == "Parameters":
 139                 # --- > Par principe, M = Id, Q = 0
 140                 Xn_predicted = Xn
 141                 Pn_predicted = Pn
 142             #
 143             if self._parameters["EstimationOf"] == "State":
 144                 d  = Ynpu - Ht * Xn_predicted
 145             elif self._parameters["EstimationOf"] == "Parameters":
 146                 d  = Ynpu - Ht * Xn_predicted
 147                 if Cm is not None and Un is not None: # Attention : si Cm est aussi dans H, doublon !
 148                     d = d - Cm * Un
 149             #
 150             _A = R + numpy.dot(Ht, Pn_predicted * Ha)
 151             _u = numpy.linalg.solve( _A , d )
 152             Xn = Xn_predicted + Pn_predicted * Ha * _u
 153             Kn = Pn_predicted * Ha * (R + numpy.dot(Ht, Pn_predicted * Ha)).I
 154             Pn = Pn_predicted - Kn * Ht * Pn_predicted
 155             #
 156             self.StoredVariables["Analysis"].store( Xn.A1 )
 157             if self._toStore("APosterioriCovariance"):
 158                 self.StoredVariables["APosterioriCovariance"].store( Pn )
 159             if self._toStore("Innovation"):
 160                 self.StoredVariables["Innovation"].store( numpy.ravel( d.A1 ) )
 161             if self._parameters["StoreInternalVariables"] \
 162                 or self._toStore("CurrentState"):
 163                 self.StoredVariables["CurrentState"].store( Xn )
 164             if self._parameters["StoreInternalVariables"] \
 165                 or self._toStore("PredictedState"):
 166                 self.StoredVariables["PredictedState"].store( Xn_predicted )
 167             if self._parameters["StoreInternalVariables"] \
 168                 or self._toStore("CostFunctionJ") \
 169                 or self._toStore("CostFunctionJb") \
 170                 or self._toStore("CostFunctionJo"):
 171                 Jb  = 0.5 * (Xn - Xb).T * BI * (Xn - Xb)
 172                 Jo  = 0.5 * d.T * RI * d
 173                 J   = float( Jb ) + float( Jo )
 174                 self.StoredVariables["CostFunctionJb"].store( Jb )
 175                 self.StoredVariables["CostFunctionJo"].store( Jo )
 176                 self.StoredVariables["CostFunctionJ" ].store( J )
 177                 if J < previousJMinimum:
 178                     previousJMinimum  = J
 179                     Xa                = Xn
 180                     if self._toStore("APosterioriCovariance"):
 181                         covarianceXa  = Pn
 182             else:
 183                 Xa = Xn
 184             #
 185         #
 186         # Stockage supplementaire de l'optimum en estimation de parametres
 187         # ----------------------------------------------------------------
 188         if self._parameters["EstimationOf"] == "Parameters":
 189             self.StoredVariables["Analysis"].store( Xa.A1 )
 190             if self._toStore("APosterioriCovariance"):
 191                 self.StoredVariables["APosterioriCovariance"].store( covarianceXa )
 192         #
 193         if self._toStore("BMA"):
 194             self.StoredVariables["BMA"].store( numpy.ravel(Xb) - numpy.ravel(Xa) )
 195         #
 196         self._post_run(HO)
 197         return 0
 198
 199 # ==============================================================================
 200 if __name__ == "__main__":
 201     print('\n AUTODIAGNOSTIC \n')