estimation of markov parameter by recursive least square method

I have to estimate Markov parameter recursively. Markov Parameters are defined by h0 = D, hi = C*Ae(i-1)*B , A,B,C,D are state-space matrices.

I have input signal u = idinput(8191,'prbs'), Gs = 2/s2+3s+2 Genertae Output Y = lsim(G,u,t)

Now How can I estimate Markov Parameter????????

Answer 1

[Rajiv Singh]

Would something like this work? z = iddata(Y,u, t(2)-t(1)); m = n4sid(z,2); h = impulse(m)

[Nehul Thakkar]

Thx for yr answer....

what mean by t(2)-t(1)

Answer 2

[k k d]

% G(s)=(s+4)/(s^3+6s^2+11s+6) Example from 'Modeling and identification of

% dynamic sysytem' by Sinha and Kuszta [Page-142]

clc

clear all

s=tf('s');

num=[10 50 200 400]; % Enter numerater coefficients from highest power of s

den=[1 33 292 660 400]; % Enter denomerater coefficients from highest power of s

G=tf(num,den)

[num,den]=tfdata(G,'v');

numf=fliplr(num) ; % For Time-moment

denf=fliplr(den) ; % For Time-moment

N=7 ; % N is number of parameters required

for i=1:N

numf=[numf 0]; % For Time-moment

TimeMoments=deconv(numf,denf); % For Time-moment

num=[num 0]; % For Markov Parameters

MarkovParameter=deconv(num,den); % For Markov Parameters

end

TimeMoments

MarkovParameter