src/daComposant/daAlgorithms/KalmanFilter.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # Copyright (C) 2008-2018 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  = ["APosterioriCorrelations", "APosterioriCovariance", "APosterioriStandardDeviations", "APosterioriVariances", "BMA", "CurrentState", "CostFunctionJ", "CostFunctionJb", "CostFunctionJo", "Innovation"]
  50             )
  51         self.requireInputArguments(
  52             mandatory= ("Xb", "Y", "HO", "R", "B" ),
  53             optional = ("U", "EM", "CM", "Q"),
  54             )
  55
  56     def run(self, Xb=None, Y=None, U=None, HO=None, EM=None, CM=None, R=None, B=None, Q=None, Parameters=None):
  57         self._pre_run(Parameters, Xb, Y, R, B, Q)
  58         #
  59         if self._parameters["EstimationOf"] == "Parameters":
  60             self._parameters["StoreInternalVariables"] = True
  61         #
  62         # Opérateurs
  63         # ----------
  64         Ht = HO["Tangent"].asMatrix(Xb)
  65         Ha = HO["Adjoint"].asMatrix(Xb)
  66         #
  67         if self._parameters["EstimationOf"] == "State":
  68             Mt = EM["Tangent"].asMatrix(Xb)
  69             Ma = EM["Adjoint"].asMatrix(Xb)
  70         #
  71         if CM is not None and "Tangent" in CM and U is not None:
  72             Cm = CM["Tangent"].asMatrix(Xb)
  73         else:
  74             Cm = None
  75         #
  76         # Nombre de pas identique au nombre de pas d'observations
  77         # -------------------------------------------------------
  78         if hasattr(Y,"stepnumber"):
  79             duration = Y.stepnumber()
  80         else:
  81             duration = 2
  82         #
  83         # Précalcul des inversions de B et R
  84         # ----------------------------------
  85         if self._parameters["StoreInternalVariables"]:
  86             BI = B.getI()
  87             RI = R.getI()
  88         #
  89         # Initialisation
  90         # --------------
  91         Xn = Xb
  92         Pn = B
  93         #
  94         self.StoredVariables["Analysis"].store( Xn.A1 )
  95         if "APosterioriCovariance" in self._parameters["StoreSupplementaryCalculations"]:
  96             self.StoredVariables["APosterioriCovariance"].store( Pn.asfullmatrix(Xn.size) )
  97             covarianceXa = Pn
  98         Xa               = Xn
  99         previousJMinimum = numpy.finfo(float).max
 100         #
 101         for step in range(duration-1):
 102             if hasattr(Y,"store"):
 103                 Ynpu = numpy.asmatrix(numpy.ravel( Y[step+1] )).T
 104             else:
 105                 Ynpu = numpy.asmatrix(numpy.ravel( Y )).T
 106             #
 107             if U is not None:
 108                 if hasattr(U,"store") and len(U)>1:
 109                     Un = numpy.asmatrix(numpy.ravel( U[step] )).T
 110                 elif hasattr(U,"store") and len(U)==1:
 111                     Un = numpy.asmatrix(numpy.ravel( U[0] )).T
 112                 else:
 113                     Un = numpy.asmatrix(numpy.ravel( U )).T
 114             else:
 115                 Un = None
 116             #
 117             if self._parameters["EstimationOf"] == "State":
 118                 Xn_predicted = Mt * Xn
 119                 if Cm is not None and Un is not None: # Attention : si Cm est aussi dans M, doublon !
 120                     Cm = Cm.reshape(Xn.size,Un.size) # ADAO & check shape
 121                     Xn_predicted = Xn_predicted + Cm * Un
 122                 Pn_predicted = Q + Mt * Pn * Ma
 123             elif self._parameters["EstimationOf"] == "Parameters":
 124                 # --- > Par principe, M = Id, Q = 0
 125                 Xn_predicted = Xn
 126                 Pn_predicted = Pn
 127             #
 128             if self._parameters["EstimationOf"] == "State":
 129                 d  = Ynpu - Ht * Xn_predicted
 130             elif self._parameters["EstimationOf"] == "Parameters":
 131                 d  = Ynpu - Ht * Xn_predicted
 132                 if Cm is not None and Un is not None: # Attention : si Cm est aussi dans H, doublon !
 133                     d = d - Cm * Un
 134             #
 135             _A = R + Ht * Pn_predicted * Ha
 136             _u = numpy.linalg.solve( _A , d )
 137             Xn = Xn_predicted + Pn_predicted * Ha * _u
 138             Kn = Pn_predicted * Ha * (R + Ht * Pn_predicted * Ha).I
 139             Pn = Pn_predicted - Kn * Ht * Pn_predicted
 140             #
 141             self.StoredVariables["Analysis"].store( Xn.A1 )
 142             if "APosterioriCovariance" in self._parameters["StoreSupplementaryCalculations"]:
 143                 self.StoredVariables["APosterioriCovariance"].store( Pn )
 144             if "Innovation" in self._parameters["StoreSupplementaryCalculations"]:
 145                 self.StoredVariables["Innovation"].store( numpy.ravel( d.A1 ) )
 146             if self._parameters["StoreInternalVariables"] or "CurrentState" in self._parameters["StoreSupplementaryCalculations"]:
 147                 self.StoredVariables["CurrentState"].store( Xn )
 148             if self._parameters["StoreInternalVariables"] \
 149                 or "CostFunctionJ" in self._parameters["StoreSupplementaryCalculations"] \
 150                 or "CostFunctionJb" in self._parameters["StoreSupplementaryCalculations"] \
 151                 or "CostFunctionJo" in self._parameters["StoreSupplementaryCalculations"]:
 152                 Jb  = 0.5 * (Xn - Xb).T * BI * (Xn - Xb)
 153                 Jo  = 0.5 * d.T * RI * d
 154                 J   = float( Jb ) + float( Jo )
 155                 self.StoredVariables["CostFunctionJb"].store( Jb )
 156                 self.StoredVariables["CostFunctionJo"].store( Jo )
 157                 self.StoredVariables["CostFunctionJ" ].store( J )
 158                 if J < previousJMinimum:
 159                     previousJMinimum  = J
 160                     Xa                = Xn
 161                     if "APosterioriCovariance" in self._parameters["StoreSupplementaryCalculations"]:
 162                         covarianceXa  = Pn
 163             else:
 164                 Xa = Xn
 165             #
 166         #
 167         # Stockage supplementaire de l'optimum en estimation de parametres
 168         # ----------------------------------------------------------------
 169         if self._parameters["EstimationOf"] == "Parameters":
 170             self.StoredVariables["Analysis"].store( Xa.A1 )
 171             if "APosterioriCovariance" in self._parameters["StoreSupplementaryCalculations"]:
 172                 self.StoredVariables["APosterioriCovariance"].store( covarianceXa )
 173         #
 174         if "BMA" in self._parameters["StoreSupplementaryCalculations"]:
 175             self.StoredVariables["BMA"].store( numpy.ravel(Xb) - numpy.ravel(Xa) )
 176         #
 177         self._post_run(HO)
 178         return 0
 179
 180 # ==============================================================================
 181 if __name__ == "__main__":
 182     print('\n AUTODIAGNOSTIC \n')