home *** CD-ROM | disk | FTP | other *** search
- function [thm,yhat,p,phi,psi] = rarmax(z,nn,adm,adg,th0,p0,phi,psi)
- %RARMAX Computes estimates recursively for an ARMAX model using the
- % Recursive Prediction Error Method
- %
- % [THM,YHAT] = rarmax(Z,NN,adm,adg)
- %
- % Z: The output-input data z=[y u] (single input only!)
- % NN : NN=[na nb nc nk], The orders and delay of an ARMAX
- % input-output model (see HELP ARMAX)
- % adm: Adaptation mechanism. adg: Adaptation gain
- % adm='ff', adg=lam: Forgetting factor algorithm, with forg factor lam
- % adm='kf', adg=R1: The Kalman filter algorithm with R1 as covariance
- % matrix of the parameter changes per time step
- % adm='ng', adg=gam: A normalized gradient algorithm, with gain gam
- % adm='ug', adg=gam: An Unnormalized gradient algorithm with gain gam
- % THM: The resulting estimates. Row k contains the estimates "in alpha-
- % betic order" corresponding to data up to time k (row k in Z)
- % YHAT: The predicted values of the output. Row k corresponds to time k.
- % Initial value of parameters(TH0) and of "P-matrix" (P0) can be given by
- % [THM,YHAT,P] = rarmax(Z,NN,adm,adg,TH0,P0)
- % Initial and last values of auxiliary data vectors phi and psi are
- % obtained by [THM,YHAT,P,phi,psi]=rarmax(Z,NN,adm,adg,TH0,P0,phi0,psi0).
- %
- % See also RARX, ROE, RBJ, RPEM and RPLR.
-
- % L. Ljung 10-1-89
- % Copyright (c) 1989 by the MathWorks, Inc.
- % All Rights Reserved.
-
-
- [nz,ns]=size(z);[ordnr,ordnc]=size(nn);
- if ns>2,error('This routine is for single input only. Use RPEM instead!'),end
- if ns==1, if ordnc~=2;error('For a time series nn should be [na nc]!'),end
- else if ordnc~=4, error('the argument nn should be [na nb nc nk]!'),end,end
- if ns==1,na=nn(1);nb=0;nc=nn(2);nk=1;
- else na=nn(1);nb=nn(2);nc=nn(3);nk=nn(4);nu=1;
- end
- if nk<1,error('Sorry, this routine requires nk>0; Shift input sequence if necessary!'),end
- d=na+nb+nc;
- if ns>2,error('Sorry, this routine is for single input only!'),end
- if ns==1,nb=0;end
- if nb==0,nk=1;end
- nam=max([na,nc]);nbm=max([nb+nk-1,nc]);
- tic=na+nb+1:na+nb+nc;
- ia=1:na;iac=1:nc;
- ib=nam+nk:nam+nb+nk-1;ibc=nam+1:nam+nc;
- ic=nam+nbm+1:nam+nbm+nc;
-
- iia=1:nam-1;iib=nam+1:nam+nbm-1;iic=nam+nbm+1:nam+nbm+nc-1;
- dm=nam+nbm+nc; if nb==0,iib=[];end
- ii=[iia iib iic];i=[ia ib ic];
-
- if nargin<8, psi=zeros(dm,1);end
- if nargin<7, phi=zeros(dm,1);end
- if nargin<6, p0=10000*eye(d);end
- if nargin<5, th0=eps*ones(d,1);end
- if isempty(psi),psi=zeros(dm,1);end
- if isempty(phi),phi=zeros(dm,1);end
- if isempty(p0),p0=10000*eye(d);end
- if isempty(th0),th0=eps*ones(d,1);end
- if length(th0)~=d, error('The length of th0 must equal the number of estimated parameters!'),end
- [th0nr,th0nc]=size(th0);if th0nr<th0nc, th0=th0';end
- p=p0;th=th0;
- if adm(1)=='f', R1=zeros(d,d);lam=adg;end
- if adm(1)=='k', [sR1,SR1]=size(adg);
- if sR1~=d | SR1~=d,error('The R1 matrix should be a square matrix with dimension equal to number of parameters!'),end
- R1=adg;lam=1;
- end
- if adm(2)=='g', grad=1;else grad=0;end
-
- for kcou=1:nz
- yh=phi(i)'*th;
- epsi=z(kcou,1)-yh;
- if ~grad,K=p*psi(i)/(lam + psi(i)'*p*psi(i));
- p=(p-K*psi(i)'*p)/lam+R1;
- else K=adg*psi(i);end
- if adm(1)=='n', K=K/(eps+psi(i)'*psi(i));end
- th=th+K*epsi;
- if nc>0,c=fstab([1;th(tic)])';else c=1;end
- th(tic)=c(2:nc+1);
- epsilon=z(kcou,1)-phi(i)'*th;
- if nb>0,zb=[z(kcou,2),-psi(ibc)'];else zb=[];end
- ztil=[[z(kcou,1),psi(iac)'];zb;[epsilon,-psi(ic)']]*c;
-
- phi(ii+1)=phi(ii);psi(ii+1)=psi(ii);
- if na>0,phi(1)=-z(kcou,1);psi(1)=-ztil(1);end
- if nb>0,phi(nam+1)=z(kcou,2);psi(nam+1)=ztil(2);end
- if nb==0,zc=ztil(2);else zc=ztil(3);end
- if nc>0,phi(nam+nbm+1)=epsilon;psi(nam+nbm+1)=zc;end
-
- thm(kcou,:)=th';yhat(kcou)=yh;
- end
- yhat=yhat';