src/daComposant/daAlgorithms/KalmanFilter.py

   1 #-*-coding:iso-8859-1-*-
   2 #
   3 # Copyright (C) 2008-2017 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
  52     def run(self, Xb=None, Y=None, U=None, HO=None, EM=None, CM=None, R=None, B=None, Q=None, Parameters=None):
  53         self._pre_run(Parameters)
  54         #
  55         if self._parameters["EstimationOf"] == "Parameters":
  56             self._parameters["StoreInternalVariables"] = True
  57         #
  58         # Opérateurs
  59         # ----------
  60         if B is None:
  61             raise ValueError("Background error covariance matrix has to be properly defined!")
  62         if R is None:
  63             raise ValueError("Observation error covariance matrix has to be properly defined!")
  64         #
  65         Ht = HO["Tangent"].asMatrix(Xb)
  66         Ha = HO["Adjoint"].asMatrix(Xb)
  67         #
  68         if self._parameters["EstimationOf"] == "State":
  69             Mt = EM["Tangent"].asMatrix(Xb)
  70             Ma = EM["Adjoint"].asMatrix(Xb)
  71         #
  72         if CM is not None and CM.has_key("Tangent") and U is not None:
  73             Cm = CM["Tangent"].asMatrix(Xb)
  74         else:
  75             Cm = None
  76         #
  77         # Nombre de pas identique au nombre de pas d'observations
  78         # -------------------------------------------------------
  79         if hasattr(Y,"stepnumber"):
  80             duration = Y.stepnumber()
  81         else:
  82             duration = 2
  83         #
  84         # Précalcul des inversions de B et R
  85         # ----------------------------------
  86         if self._parameters["StoreInternalVariables"]:
  87             BI = B.getI()
  88             RI = R.getI()
  89         #
  90         # Initialisation
  91         # --------------
  92         Xn = Xb
  93         Pn = B
  94         #
  95         self.StoredVariables["Analysis"].store( Xn.A1 )
  96         if "APosterioriCovariance" in self._parameters["StoreSupplementaryCalculations"]:
  97             self.StoredVariables["APosterioriCovariance"].store( Pn.asfullmatrix(Xn.size) )
  98             covarianceXa = Pn
  99         Xa               = Xn
 100         previousJMinimum = numpy.finfo(float).max
 101         #
 102         for step in range(duration-1):
 103             if hasattr(Y,"store"):
 104                 Ynpu = numpy.asmatrix(numpy.ravel( Y[step+1] )).T
 105             else:
 106                 Ynpu = numpy.asmatrix(numpy.ravel( Y )).T
 107             #
 108             if U is not None:
 109                 if hasattr(U,"store") and len(U)>1:
 110                     Un = numpy.asmatrix(numpy.ravel( U[step] )).T
 111                 elif hasattr(U,"store") and len(U)==1:
 112                     Un = numpy.asmatrix(numpy.ravel( U[0] )).T
 113                 else:
 114                     Un = numpy.asmatrix(numpy.ravel( U )).T
 115             else:
 116                 Un = None
 117             #
 118             if self._parameters["EstimationOf"] == "State":
 119                 Xn_predicted = Mt * Xn
 120                 if Cm is not None and Un is not None: # Attention : si Cm est aussi dans M, doublon !
 121                     Cm = Cm.reshape(Xn.size,Un.size) # ADAO & check shape
 122                     Xn_predicted = Xn_predicted + Cm * Un
 123                 Pn_predicted = Q + Mt * Pn * Ma
 124             elif self._parameters["EstimationOf"] == "Parameters":
 125                 # --- > Par principe, M = Id, Q = 0
 126                 Xn_predicted = Xn
 127                 Pn_predicted = Pn
 128             #
 129             if self._parameters["EstimationOf"] == "State":
 130                 d  = Ynpu - Ht * Xn_predicted
 131             elif self._parameters["EstimationOf"] == "Parameters":
 132                 d  = Ynpu - Ht * Xn_predicted
 133                 if Cm is not None and Un is not None: # Attention : si Cm est aussi dans H, doublon !
 134                     d = d - Cm * Un
 135             #
 136             _A = R + Ht * Pn_predicted * Ha
 137             _u = numpy.linalg.solve( _A , d )
 138             Xn = Xn_predicted + Pn_predicted * Ha * _u
 139             Kn = Pn_predicted * Ha * (R + Ht * Pn_predicted * Ha).I
 140             Pn = Pn_predicted - Kn * Ht * Pn_predicted
 141             #
 142             self.StoredVariables["Analysis"].store( Xn.A1 )
 143             if "APosterioriCovariance" in self._parameters["StoreSupplementaryCalculations"]:
 144                 self.StoredVariables["APosterioriCovariance"].store( Pn )
 145             if "Innovation" in self._parameters["StoreSupplementaryCalculations"]:
 146                 self.StoredVariables["Innovation"].store( numpy.ravel( d.A1 ) )
 147             if self._parameters["StoreInternalVariables"]:
 148                 Jb  = 0.5 * (Xn - Xb).T * BI * (Xn - Xb)
 149                 Jo  = 0.5 * d.T * RI * d
 150                 J   = float( Jb ) + float( Jo )
 151                 if self._parameters["StoreInternalVariables"] or "CurrentState" in self._parameters["StoreSupplementaryCalculations"]:
 152                     self.StoredVariables["CurrentState"].store( Xn )
 153                 self.StoredVariables["CostFunctionJb"].store( Jb )
 154                 self.StoredVariables["CostFunctionJo"].store( Jo )
 155                 self.StoredVariables["CostFunctionJ" ].store( J )
 156                 if J < previousJMinimum:
 157                     previousJMinimum  = J
 158                     Xa                = Xn
 159                     if "APosterioriCovariance" in self._parameters["StoreSupplementaryCalculations"]:
 160                         covarianceXa  = Pn
 161             else:
 162                 Xa = Xn
 163             #
 164         #
 165         # Stockage supplementaire de l'optimum en estimation de parametres
 166         # ----------------------------------------------------------------
 167         if self._parameters["EstimationOf"] == "Parameters":
 168             self.StoredVariables["Analysis"].store( Xa.A1 )
 169             if "APosterioriCovariance" in self._parameters["StoreSupplementaryCalculations"]:
 170                 self.StoredVariables["APosterioriCovariance"].store( covarianceXa )
 171         #
 172         if "BMA" in self._parameters["StoreSupplementaryCalculations"]:
 173             self.StoredVariables["BMA"].store( numpy.ravel(Xb) - numpy.ravel(Xa) )
 174         #
 175         self._post_run(HO)
 176         return 0
 177
 178 # ==============================================================================
 179 if __name__ == "__main__":
 180     print '\n AUTODIAGNOSTIC \n'