www.pudn.com > programradarsystemdisign.zip > fig9_27.m

clear all 
npts = 2000; 
del = 1/2000; 
t = 0:del:1; 
inp = (1+.2 .* t + .1 .*t.^2);% + cos(2. * pi * 2.5 .* t); 
X0 = [1,.1,.01]'; 
% it is assumed that the measurmeny vector H=[1,0,0] 
% this is the update interval in seconds 
T = 1.; 
% enter the mesurement noise variance 
R = .01; 
% this is the state noise variance 
nvar = .18; 
[residual, estimate] = kalman_filter(npts, T, X0, inp, R, nvar); 
figure(1) 
plot(residual(1:500),'k') 
xlabel ('Sample number') 
ylabel ('Residual') 
figure(2) 
subplot(2,1,1) 
plot(inp,'k') 
axis tight 
ylabel ('position - truth') 
subplot(2,1,2) 
plot(estimate,'k') 
axis tight 
xlabel ('Sample number') 
ylabel ('Predicted position')