Mercurial > hg > ltpda

comparison m-toolbox/test/utils/test_2dim_vcfit.m @ 0:f0afece42f48

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Import.
author Daniele Nicolodi <nicolodi@science.unitn.it>
date Wed, 23 Nov 2011 19:22:13 +0100
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:f0afece42f48
1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2 % TEST Fitting procedure in 2dim z-domain VDFIT
3 % At the end you have 4 stable transfer functions in partial fractions
4 %
5 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
6 % HISTORY: 02-10-2008 L Ferraioli
7 % Creation
8 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
9 %% VERSION
10
11 '$Id: test_2dim_vcfit.m,v 1.1 2009/04/23 10:11:26 luigi Exp $';
12
13 %% Clear
14
15 clear all
16
17 %% Loading spectra
18
19 load ..\m-toolbox\test\mpsd.mat % load mpsd.mat first column is f then psd1, csd and psd2
20
21 f = mpsd(:,1);
22 psd1 = mpsd(:,2);
23 csd = mpsd(:,3);
24 psd2 = mpsd(:,4);
25 fs = 10;
26
27
28 %% Eigendecomposition
29
30 [tf11,tf12,tf21,tf22] = utils.math.eigcsd(psd1,csd,conj(csd),psd2);
31
32 %% Constructing vector
33
34 f1 = [tf11 tf21];
35 f2 = [tf12 tf22];
36
37 %% VCFIT
38
39 N = 26; %Order of approximation
40
41
42 % Max Iteration
43 Nmaxiter = 70;
44
45 % Fitting params
46 fitin.stable = 0;
47 fitin.dterm = 0;
48 fitin.plot = 0;
49
50 weight = utils.math.wfun(f1,2);
51 % weight = 1./abs(f1);
52
53 pparams = struct('spolesopt',2, 'type','CONT', 'pamp', 0.01);
54 poles1 = utils.math.startpoles(N,f,pparams);
55 for hh = 1:Nmaxiter
56 [res1,poles1,dterm1,mresp1,rdl1,mse1] = utils.math.vcfit(f1,f,poles1,weight,fitin); % Fitting
57 % disp(['Iter' num2str(hh)])
58
59 % Stop condition checking
60 order = length(poles1);
61 mlr1(hh) = mse1(:,1);
62 mlr2(hh) = mse1(:,2);
63 % [ext1,msg1] = stopfit(f1(:,1),rdl1(:,1),mlr1,order,'mlsrvar',2,0.0001,2);
64 % [ext2,msg2] = stopfit(f1(:,2),rdl1(:,2),mlr2,order,'mlsrvar',2,0.0001,2);
65 %
66 % if ext1 && ext2
67 % disp(msg1)
68 % break
69 % end
70
71 end
72
73 % plotting squared error
74 figure()
75 semilogy(mlr1,'-ok')
76 hold on
77 grid on
78 semilogy(mlr2,'-or')
79
80 % % plotting RMS error variation
81 % figure()
82 % semilogy(abs(diff(sqrt(mlr1))),'-ok')
83 % hold on
84 % grid on
85 % semilogy(abs(diff(sqrt(mlr2))),'-or')
86
87 % plotting squared error variation
88 figure()
89 semilogy(abs(diff(mlr1)./mlr1(1,end-1)),'-ok')
90 hold on
91 grid on
92 semilogy(abs(diff(mlr2)./mlr2(1,end-1)),'-or')
93
94 %% All Passing
95
96 [np1,resp1] = pfallps(res1,poles1,dterm1,mresp1,f);
97
98
99 figure()
100 subplot(2,1,1);
101 p1 = loglog(f,abs(mresp1),'k');
102 hold on
103 p2 = loglog(f,abs(resp1),'r');
104 xlabel('Frequency [Hz]')
105 ylabel('Amplitude')
106 legend([p1(1) p2(1)],'VDFIT','Stabilized')
107 hold off
108
109 subplot(2,1,2);
110 p4 = semilogx(f,(180/pi).*unwrap(angle(mresp1)),'k');
111 hold on
112 p5 = semilogx(f,(180/pi).*unwrap(angle(resp1)),'r');
113 xlabel('Frequency [Hz]')
114 ylabel('Phase [Deg]')
115 legend([p4(1) p5(1)],'VDFIT', 'Stabilized')
116 hold off
117
118 %% VCFIT
119
120 N = 14; %Order of approximation
121
122
123 % Max Iteration
124 Nmaxiter = 50;
125
126 % Fitting params
127 fitin.stable = 0;
128 fitin.dterm = 0;
129 fitin.plot = 1;
130
131 weight = wfun(f2,1);
132
133 pparams = struct('spolesopt',1, 'type','CONT', 'pamp', 0.01);
134 poles2 = startpoles(N,f,pparams);
135
136 for hh = 1:Nmaxiter
137 [res2,poles2,dterm2,mresp2,rdl2] = vcfit(f2,f,poles2,weight,fitin); % Fitting
138 disp(['Iter' num2str(hh)])
139
140 % Stop condition checking
141 order = length(poles1);
142 mlr1(hh) = mean(log10(abs(f2(:,1)))-log10(abs(rdl2(:,1))));
143 mlr2(hh) = mean(log10(abs(f2(:,2)))-log10(abs(rdl2(:,2))));
144 [ext1,msg1] = stopfit(f2(:,1),rdl2(:,1),mlr1,order,'mlrsvar',2,0.0001,2);
145 [ext2,msg2] = stopfit(f2(:,2),rdl2(:,2),mlr2,order,'mlrsvar',2,0.0001,2);
146
147 if ext1 && ext2
148 disp(msg1)
149 break
150 end
151
152 end
153