root/pmsm/mpf_test.cpp @ 229

/*!
  \file
  \brief TR 2525 file for testing Toy Problem of mpf for Covariance Estimation
  \author Vaclav Smidl.

  -----------------------------------
  BDM++ - C++ library for Bayesian Decision Making under Uncertainty

  Using IT++ for numerical operations
  -----------------------------------
*/


#include <itpp/itbase.h>
#include <estim/libKF.h>
#include <estim/libPF.h>
#include <estim/ekf_templ.h>
#include <stat/libFN.h>

#include "pmsm.h"
#include "simulator.h"
#include "sim_profiles.h"

using namespace itpp;

int main() {
        // Kalman filter
        int Ndat = 9000;
        double h = 1e-6;
        int Nsimstep = 125;
        int Npart = 20;

        // internal model
        IMpmsm fxu;
        //                  Rs    Ls        dt       Fmag(Ypm)    kp   p    J     Bf(Mz)
        fxu.set_parameters ( 0.28, 0.003465, Nsimstep*h, 0.1989, 1.5 ,4.0, 0.04, 0.0 );
        // observation model
        OMpmsm hxu;

        vec mu0= "0.0 0.0 0.0 0.0";
        vec Qdiag ( "1e-6 1e-6 0.001 0.0001" ); //zdenek: 0.01 0.01 0.0001 0.0001
        vec Rdiag ( "1e-8 1e-8" ); //var(diff(xth)) = "0.034 0.034"
        chmat Q ( Qdiag );
        chmat R ( Rdiag );
        EKFCh KFE ( rx,ry,ru );
        KFE.set_parameters ( &fxu,&hxu,Q,R );
        KFE.set_est ( mu0, chmat ( zeros ( 4 ) ) );

        RV rQ ( "{Q }","4" );
        EKFCh_unQ KFEp ( rx,ry,ru,rQ );
        KFEp.set_parameters ( &fxu,&hxu,Q,R );
        KFEp.set_est ( mu0, chmat ( zeros ( 4 ) ) );

        //mgamma_fix evolQ ( rQ,rQ );
        migamma_fix evolQ ( rQ,rQ );
        MPF<EKFCh_unQ> M ( rx,rQ,evolQ,evolQ,Npart,KFEp );
        // initialize
        evolQ.set_parameters ( 0.1, Qdiag, 1.0); //sigma = 1/10 mu
        evolQ.condition (Qdiag ); //Zdenek default
        double xxx;
        cout << Qdiag <<endl << "smp:"<< evolQ.samplecond(Qdiag,xxx) <<endl;
        eigamma* pfinit=dynamic_cast<eigamma*>(evolQ._e());
        cout << pfinit->mean()<<endl;
        M.set_est ( *pfinit );
        evolQ.set_parameters ( 0.10, Qdiag,0.995); //sigma = 1/10 mu
        cout << Qdiag <<endl << "smp:"<< evolQ.samplecond(Qdiag,xxx) <<endl;
        
        //

        const epdf& KFEep = KFE._epdf();
        const epdf& Mep = M._epdf();

        mat Xt=zeros ( Ndat ,4 ); //true state from simulator
        mat Dt=zeros ( Ndat,2+2 ); //observation
        mat XtE=zeros ( Ndat, 4 );
        mat VarE=zeros ( Ndat, 4 );
        mat Qtr=zeros ( Ndat, 4 );
        mat XtM=zeros ( Ndat,4+4 ); //Q + x
        mat VarM=zeros ( Ndat,4+4 ); //Q + x

        // SET SIMULATOR
        pmsmsim_set_parameters ( 0.28,0.003465,0.1989,0.0,4,1.5,0.04, 200., 3e-6, h );
        vec dt ( 2 );
        vec ut ( 2 );
        vec xt ( 4 );
        vec xtm=zeros(4);
        double Ww=0.0;
        vec vecW="1 2 4 9 4 2 0 -4 -9 -16 -4 0 0";

        for ( int tK=1;tK<Ndat;tK++ ) {
                //Number of steps of a simulator for one step of Kalman
                for ( int ii=0; ii<Nsimstep;ii++ ) {
                        //simulator
                        sim_profile_vec01t(Ww,vecW);
                        pmsmsim_step ( Ww );
                };
                ut(0) = KalmanObs[4];
                ut(1) = KalmanObs[5];
                xt = fxu.eval(xtm,ut) + diag(sqrt(Qdiag))*randn(4);
                dt = hxu.eval(xt,ut);
                xtm = xt;
                
                
                //Variances 
                if (tK==1000)  Qdiag(0)*=10; 
                if (tK==2000) Qdiag(0)/=10; 
                if (tK==3000)  Qdiag(1)*=10; 
                if (tK==4000) Qdiag(1)/=10; 
                if (tK==5000)  Qdiag(2)*=10; 
                if (tK==6000) Qdiag(2)/=10; 
                if (tK==7000)  Qdiag(3)*=10; 
                if (tK==8000) Qdiag(3)/=10; 
                
                //estimator
                KFE.bayes ( concat ( dt,ut ) );
                M.bayes ( concat ( dt,ut ) );

                Xt.set_row ( tK, xt); //vec from C-array
                Dt.set_row ( tK, concat ( dt,ut));
                Qtr.set_row ( tK, Qdiag);
                XtE.set_row ( tK,KFEep.mean() );
                VarE.set_row ( tK,KFEep.variance() );
                XtM.set_row ( tK,Mep.mean() );
                VarM.set_row ( tK,Mep.variance() );
        }

        it_file fou ( "mpf_test.it" );

        fou << Name ( "xth" ) << Xt;
        fou << Name ( "Dt" ) << Dt;
        fou << Name ( "Qtr" ) << Qtr;
        fou << Name ( "xthE" ) << XtE;
        fou << Name ( "xthM" ) << XtM;
        fou << Name ( "VarE" ) << VarE;
        fou << Name ( "VarM" ) << VarM;
        //Exit program:

        return 0;
}