Mercurial > hg > ltpda

comparison m-toolbox/test/MDC2/test_ao_noisegen2D_mdc2.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 % A test script for ao/noisegen2D
2 %
3 % DESCRIPTION: Run noisegen2D with mdc2 models and test procedure accuracy
4 %
5 % L. Ferraioli 04-02-2009
6 %
7 % $Id: test_ao_noisegen2D_mdc2.m,v 1.1 2009/04/20 14:31:43 luigi Exp $
8 %
9
10 %% General use variables and vectors
11
12 f = logspace(-6,log10(5),300);
13 f = f.';
14 fs = 10;
15 Nsecs = 1e5; % number of seconds
16 Nfft = 1e5; % number of samples for the fft
17 pls = plist('Nfft', Nfft,'Order',0); % plist for spectra
18
19 %% MDC2 Models
20
21 b = ao(plist('built-in','mdc2r2_fd_ltpnoise','f',f,'fs',fs));
22 CSD = [b(1) b(2);conj(b(2)) b(3)];
23
24 %% Make white noise
25
26 a1 = ao(plist('tsfcn', 'randn(size(t))', 'fs', fs, 'nsecs', Nsecs));
27 a2 = ao(plist('tsfcn', 'randn(size(t))', 'fs', fs, 'nsecs', Nsecs));
28 a1.setYunits('m');
29 a2.setYunits('m');
30 a = [a1 a2];
31
32 % axx = a.cpsd(pls);
33
34 %%
35 iplot(a)
36
37 %% some ploting
38
39 iplot(axx)
40
41 iplot(CSD(1,1),CSD(1,2),CSD(2,1),CSD(2,2))
42
43 %% Noise generation
44
45 pl = plist(...
46 'csd11', CSD(1,1), ...
47 'csd12', CSD(1,2), ...
48 'csd21', CSD(2,1), ...
49 'csd22', CSD(2,2), ...
50 'MaxIter', 80, ...
51 'PoleType', 2, ...
52 'MinOrder', 30, ...
53 'MaxOrder', 40, ...
54 'Weights', 2, ...
55 'Plot', false,...
56 'FITTOL', 1e-5,...
57 'MSEVARTOL', 1e-2,...
58 'UseSym', 0,...
59 'Disp', false);
60
61 ac = noisegen2D(a, pl);
62
63 %% Checking results and starting data
64
65 % iplot(a)
66 iplot(ac)
67
68 %% Making cross-spectrum
69
70 acxx1 = ac(1).psd(pls);
71 acxx2 = ac(2).psd(pls);
72 acxx12 = cpsd(ac(1),ac(2),pls);
73 acch = ac.cohere(pls);
74
75 %% Plotting spectra
76
77 % iplot(acxx);
78
79 iplot(abs(acxx1),abs(CSD(1,1))) % model data need to be multiplied by 2 because acxx is the onesided cpsd
80 iplot(abs(acxx12),abs(CSD(1,2)))
81 iplot(abs(acxx2),abs(CSD(2,2)))
82
83 iplot(abs(acxx1),abs(acxx12),abs(acxx2))
84
85 %% Plotting coherence
86
87 iplot(acch,(abs(CSD(1,2)).^2)./(CSD(1,1).*CSD(2,2)))
88
89 %%
90
91 m1=mean(acxx(2,2).data.y(end-10,end)) % calculate average on the tail of channel 2
92 m2=mean(CSD(2,2).data.y(end-5,end))% calculate average on the tail of channel 2
93 m1/m2 % verify that the ratio is near 1
94
95 %%
96 % ************************************************************************
97 % Some more analysis for testing the accuracy of noise generation procedure
98 % ************************************************************************
99
100 %% Extracting filters from data
101
102 Filt11 = find(ac(1).procinfo,'Filt11');
103 Filt12 = find(ac(1).procinfo,'Filt12');
104 Filt21 = find(ac(2).procinfo,'Filt21');
105 Filt22 = find(ac(2).procinfo,'Filt22');
106
107 %% Calculating filters responses
108
109 tr11 = resp(Filt11,plist('f',f));
110 rFilt11 = tr11(1);
111 for ii = 2:numel(tr11)
112 rFilt11 = rFilt11 + tr11(ii);
113 end
114 rFilt11.setName('rFilt11', 'internal');
115
116 tr12 = resp(Filt12,plist('f',f));
117 rFilt12 = tr12(1);
118 for ii = 2:numel(tr12)
119 rFilt12 = rFilt12 + tr12(ii);
120 end
121 rFilt12.setName('rFilt12', 'internal');
122
123 tr21 = resp(Filt21,plist('f',f));
124 rFilt21 = tr21(1);
125 for ii = 2:numel(tr21)
126 rFilt21 = rFilt21 + tr21(ii);
127 end
128 rFilt21.setName('rFilt21', 'internal');
129
130 tr22 = resp(Filt22,plist('f',f));
131 rFilt22 = tr22(1);
132 for ii = 2:numel(tr22)
133 rFilt22 = rFilt22 + tr22(ii);
134 end
135 rFilt22.setName('rFilt22', 'internal');
136
137 %% Obtaining transfer functions
138
139 % calculating transfer functions from data
140 pl = plist('Nfft', Nt);
141 etf11 = tfe(a(1),ac(1),pl);
142 etf12 = tfe(a(2),ac(1),pl);
143 etf21 = tfe(a(1),ac(2),pl);
144 etf22 = tfe(a(2),ac(2),pl);
145
146 %% Comparing Filters Responses with estimated TFs (e-TFs)
147
148 % Comparing filters responses and calculated TFs
149 pl = plist('Legends', {'Filter Response','e-TF'});
150 iplot(rFilt11,etf11,pl)
151 iplot(rFilt12,etf12,pl)
152 iplot(rFilt21,etf21,pl)
153 iplot(rFilt22,etf22,pl)
154
155 %% Filtering data separately
156
157 % This operation is performed internally to the noisegen2D. Output data are
158 % then obtained by b1 = b11 + b12 and b2 = b21 + b22
159 b11 = filter(a1,plist('filter',Filt11,'bank','parallel'));
160 b12 = filter(a2,plist('filter',Filt12,'bank','parallel'));
161 b21 = filter(a1,plist('filter',Filt21,'bank','parallel'));
162 b22 = filter(a2,plist('filter',Filt22,'bank','parallel'));
163
164 %% Extracting transfer functions from separately filtered data se-TFs
165
166 etf11 = tfe(a1,b11,pls);
167 etf12 = tfe(a2,b12,pls);
168 etf21 = tfe(a1,b21,pls);
169 etf22 = tfe(a2,b22,pls);
170
171 %% Comparing separately-estimated TFs (se-TFs) with filter responses
172
173 pl = plist('Legends', {'Filter Response','se-TF'});
174 iplot(rFilt11,etf11,pl)
175 iplot(rFilt12,etf12,pl)
176 iplot(rFilt21,etf21,pl)
177 iplot(rFilt22,etf22,pl)
178
179 %% Comparing filters with TFs obtained by eigendecomposition
180
181 % This function output transfer functions as they are obtained by the
182 % eigendecomposition process. i.e. before the fitting process
183
184 icsd11 = CSD(1,1).data.y*fs/2;
185 icsd12 = CSD(1,2).data.y*fs/2;
186 icsd21 = CSD(2,1).data.y*fs/2;
187 icsd22 = CSD(2,2).data.y*fs/2;
188
189 [tf11,tf12,tf21,tf22] = utils.math.eigcsd(icsd11,icsd12,icsd21,icsd22,'USESYM',0,'DIG',50,'OTP','TF');
190
191 % Making AOs
192 eigtf11 = ao(fsdata(f,tf11,fs));
193 eigtf12 = ao(fsdata(f,tf12,fs));
194 eigtf21 = ao(fsdata(f,tf21,fs));
195 eigtf22 = ao(fsdata(f,tf22,fs));
196
197 %% Comparing eig-TFs with output filters
198
199 % Compare TFs before and after the fitting process
200
201 pl = plist('Legends', {'eig-TF','Filter Response'});
202 iplot(eigtf11,rFilt11,pl)
203 iplot(eigtf12,rFilt12,pl)
204 iplot(eigtf21,rFilt21,pl)
205 iplot(eigtf22,rFilt22,pl)
206
207 %% Phase difference between eig-TFs and output filters
208
209 % checking that phase differences between TFs are preserved by the fitting
210 % process
211 eigph1 = unwrap(angle(eigtf11)-angle(eigtf21));
212 eigph1.setYunits('rad')
213 filtph1 = unwrap(angle(rFilt11)-angle(rFilt21));
214 filtph1.setYunits('rad')
215 eigph2 = unwrap(angle(eigtf22)-angle(eigtf12));
216 eigph2.setYunits('rad')
217 filtph2 = unwrap(angle(rFilt22)-angle(rFilt12));
218 filtph2.setYunits('rad')
219
220 pl = plist('Legends',{'eig-TF','Filter'},'YScales',{'All','lin'});
221 iplot(eigph1+2*pi,filtph1,pl)
222 iplot(eigph2,filtph2,pl)
223
224
225 %% Comparing eig-TFs with se-TFs
226
227 % Compare eigendecomposition results with separately estimated TFs (se-TFs)
228 pl = plist('Legends', {'eig-TF','se-TF'});
229 iplot(eigtf11,etf11(1,2),pl)
230 iplot(eigtf12,etf12(1,2),pl)
231 iplot(eigtf21,etf21(1,2),pl)
232 iplot(eigtf22,etf22(1,2),pl)
233
234 %% Phase difference between eig-TFs and se-TFs
235
236 % checking that phase differences between TFs are preserved by the fitting
237 % process also for the filtered data (se-TFs)
238 eigph1 = unwrap(angle(eigtf11)-angle(eigtf21));
239 filtph1 = unwrap(angle(etf11(1,2))-angle(etf21(1,2)));
240 eigph2 = unwrap(angle(eigtf22)-angle(eigtf12));
241 filtph2 = unwrap(angle(etf22(1,2))-angle(etf12(1,2)));
242
243 pl = plist('Legends',{'eig-TF \Delta\phi','se-TF \Delta\phi'},'YScales',{'All','lin'});
244 iplot(eigph1,filtph1,pl)
245 iplot(eigph2,filtph2,pl)
246
247 % END