Mercurial > hg > ltpda

comparison m-toolbox/test/LTPDA_training/topic5/TrainigSession_T5_Ex02.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 % Training session Topic 5 exercise 02
2 %
3 % System identification in z-domain 2
4 %
5 % 1) Load fsdata object from file
6 % 2) Fit loaded TF data with zDomainFit and fixed order
7 % 3) Compare results
8 %
9 % L FERRAIOLI 22-02-09
10 %
11 % $Id: TrainigSession_T5_Ex02.m,v 1.3 2009/02/25 18:18:45 luigi Exp $
12 %
13 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
14
15 %% 1) load fsdata
16
17 % load AO from file
18 rfilt = ao(plist('filename', 'topic5\T5_Ex02_rfilt.xml'));
19 iplot(rfilt)
20
21 %% 2) Fitting TF - fixed model order
22
23 % Loaded fsdata are the response of an order 19 partial fractioned filter.
24 % We now try to fitting the loaded filter response with zDomainFit with a
25 % fixed model order.
26 % We set Autosearch to off, in this case the function do not perform
27 % accuracy test but simply run how far maximum number of iteration is
28 % reached. Model order is fixed by minorder parameter.
29
30 plfit1 = plist('FS',10,... % Sampling frequency for the model filters
31 'AutoSearch','off',... % Automatically search for a good model
32 'StartPolesOpt','c1',... % Define the properties of the starting poles - complex distributed in the unitary circle
33 'maxiter',30,... % maximum number of iteration per model order
34 'minorder',19,... % fixed model order
35 'weightparam','abs',... % assign weights as 1./abs(data)
36 'Plot','on',... % set the plot on or off
37 'ForceStability','on',... % force to output a stable ploes model
38 'CheckProgress','off'); % display fitting progress on the command window
39
40 % Do the fit
41 fobj = zDomainFit(rfilt,plfit1);
42 % setting input and output units for fitted model
43
44 %% 3) Compare results
45
46 % Extracting residues and poles from fit results
47 fRes = zeros(numel(fobj),1); % fit residue vector initialization
48 fPoles = zeros(numel(fobj),1); % fit poles vector initialization
49
50 % extracting data from fitted filters
51 for ii = 1:numel(fobj)
52 fRes(ii,1) = fobj(ii).a(1);
53 fPoles(ii,1) = -1*fobj(ii).b(2);
54 end
55 [fRes,idx] = sort(fRes);
56 fPoles = fPoles(idx);
57
58 % starting model residues and poles
59 mRes = [2.44554138162509e-011 - 1.79482547894083e-011i;
60 2.44554138162509e-011 + 1.79482547894083e-011i;
61 2.66402334803101e-009 + 1.1025122049153e-009i;
62 2.66402334803101e-009 - 1.1025122049153e-009i;
63 -7.3560293387644e-009;
64 -1.82811618589835e-009 - 1.21803627800855e-009i;
65 -1.82811618589835e-009 + 1.21803627800855e-009i;
66 1.16258677367555e-009;
67 1.65216557639319e-016;
68 -1.78092396888606e-016;
69 -2.80420398962379e-017;
70 9.21305973049041e-013 - 8.24686706827269e-014i;
71 9.21305973049041e-013 + 8.24686706827269e-014i;
72 5.10730060739905e-010 - 3.76571756625722e-011i;
73 5.10730060739905e-010 + 3.76571756625722e-011i;
74 3.45893698149735e-009;
75 3.98139182134446e-014 - 8.25503935419059e-014i;
76 3.98139182134446e-014 + 8.25503935419059e-014i;
77 -1.40595719147164e-011];
78 [mRes,idx] = sort(mRes);
79
80 mPoles = [0.843464045655194 - 0.0959986292915475i;
81 0.843464045655194 + 0.0959986292915475i;
82 0.953187595424927 - 0.0190043625473383i;
83 0.953187595424927 + 0.0190043625473383i;
84 0.967176277937188;
85 0.995012027005247 - 0.00268322602801729i;
86 0.995012027005247 + 0.00268322602801729i;
87 0.996564761885673;
88 0.999999366165445;
89 0.999981722418555;
90 0.999921882627659;
91 0.999624431675213 - 0.000813407848742761i;
92 0.999624431675213 + 0.000813407848742761i;
93 0.997312006278751 - 0.00265611346834941i;
94 0.997312006278751 + 0.00265611346834941i;
95 0.990516544257531;
96 0.477796923118318 - 0.311064085401834i;
97 0.477796923118318 + 0.311064085401834i;
98 0];
99 mPoles = mPoles(idx);
100
101 % Check the relative difference
102 (mRes-fRes)./abs(mRes)
103 (mPoles-fPoles)./abs(mPoles)
104 % Results are accurate to the 7th decimal digit
105