home *** CD-ROM | disk | FTP | other *** search
- function [thm,yhat,p,phi] = rarx(z,nn,adm,adg,th0,p0,phi)
- %RARX Computes estimates recursively for an ARX model using the
- % Recursive Least Squares Method
- %
- % [THM,YHAT] = rarx(Z,NN,adm,adg)
- %
- % Z: The output-input data z=[y u]
- % NN : NN=[na nb nk], The orders and delay of an ARX model (see HELP ARX)
- % 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 outputs. Row k corresponds to time k.
- % Initial value of parameters(TH0) and of "P-matrix" (P0) can be given by
- % [THM,YHAT,P] = rarx(Z,NN,adm,adg,TH0,P0)
- % Initial and last values of auxiliary data vector phi are
- % obtained by [THM,YHAT,P,phi]=rarx(Z,NN,adm,adg,TH0,P0,phi0).
- %
- % See also RARMAX, ROE, RBJ, RPEM and RPLR.
-
- % L. Ljung 10-1-89
- % Copyright (c) 1989-90 by the MathWorks, Inc.
- % All Rights Reserved.
-
-
- [nz,ns]=size(z);
- if ns==1,if length(nn)~=1,error('For a time series nn should be a scalar nn=na!'),end,end
- if 2*ns-1~=length(nn),error('Incorrect number of orders specified in nn. nn=[na nb nk]'),end
- na=nn(1);if ns>1,nb=nn(2:ns);nk=nn(ns+1:2*ns-1);else nk=1;nb=0;end
- nu=1;
- if any(nk<1),error('Sorry, this routine requires nk>0; Shift input sequence if necessary!'),end
- d=na+sum(nb);
- nbm=nb+nk-1;ncbm=na+cumsum([0 nbm]);
- ii=[1:na+sum(nbm)];
- i=[1:na];
- for ku=1:ns-1,i=[i ncbm(ku)+nk(ku):ncbm(ku+1)];end
-
- dm=na+sum(nbm);
-
- 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(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*phi(i)/(lam + phi(i)'*p*phi(i));
- p=(p-K*phi(i)'*p)/lam+R1;
- else K=adg*phi(i);end
- if adm(1)=='n', K=K/(eps+phi(i)'*phi(i));end
- th=th+K*epsi;
-
- epsilon=z(kcou,1)-th'*phi(i);
-
- phi(ii+1)=phi(ii);
- if na>0,phi(1)=-z(kcou,1);end
- if any(ncbm>0),phi(ncbm(1:ns-1)+1)=z(kcou,2:ns)';end
-
- thm(kcou,:)=th';yhat(kcou)=yh;
- end
- yhat=yhat';