Mercurial > hg > ltpda
view m-toolbox/classes/@ao/quasiSweptSine.m @ 52:daf4eab1a51e database-connection-manager tip
Fix. Default password should be [] not an empty string
author | Daniele Nicolodi <nicolodi@science.unitn.it> |
---|---|
date | Wed, 07 Dec 2011 17:29:47 +0100 |
parents | f0afece42f48 |
children |
line wrap: on
line source
% QUASISWEPTSING computes a transfer function from swept-sine measurements %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % DESCRIPTION: QUASISWEPTSING computes a transfer function from discrete % swept-sine measurements. % % In order for the calculation to work, you need to give it an array of % start and stop times (or durations), and (optionally) an array of % amplitudes and frequencies of the injected sine-waves. If you don't % specify the frequencies, you must give a time-series of the injected % signal and the algorithm will try to determine the amplitudes and % frequencies from the data. % % % CALL: T = quasiSweptSine(out, pl); % % INPUTS: out - The measured output of the system % PL - parameter list % % OUTPUT: T - the measured transfer function % % The procinfo of the output AOs contains the following fields: % % 'frequencies' - the frequencies used in the DFT estimation. % 'timespans' - an array of timespan objects, one for each sine-wave % segment % % % <a href="matlab:utils.helper.displayMethodInfo('ao', 'quasiSweptSine')">Parameters Description</a> % % VERSION: $Id: quasiSweptSine.m,v 1.10 2011/04/08 08:56:16 hewitson Exp $ % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% function varargout = quasiSweptSine(varargin) % check if this is a call for parameters if utils.helper.isinfocall(varargin{:}) varargout{1} = getInfo(varargin{3}); return end % tell the system we are runing import utils.const.* utils.helper.msg(msg.PROC3, 'running %s/%s', mfilename('class'), mfilename); % collect input variable names in_names = cell(size(varargin)); for ii = 1:nargin,in_names{ii} = inputname(ii);end % collect all AOs and plists [as, ao_invars] = utils.helper.collect_objects(varargin(:), 'ao', in_names); pl = utils.helper.collect_objects(varargin(:), 'plist', in_names); if nargout == 0 error('### quasiSweptSine can not be used as a modifier method. Please give at least one output'); end % Make copies or handles to inputs bs = copy(as, nargout); % combine plists pl = parse(pl, getDefaultPlist()); % Parameters input = pl.find('input'); startTimes = pl.find('Start times'); stopTimes = pl.find('Stop times'); durations = pl.find('durations'); amplitudes = pl.find('amplitudes'); frequencies = pl.find('frequencies'); % phases = pl.find('phases'); inUnits = pl.find('Input Units'); win = pl.find('Win'); Nerror = pl.find('Nerror'); if isa(input, 'ao') && ~isa(input.data, 'tsdata') utils.helper.err('quasiSweptSine requires time-series as input'); end %--------------------------------------------------- %--------- Convert the times to timespans if isempty(durations) && isempty(stopTimes) utils.helper.err('You need to specify either an array of stop times, or an array of durations'); end %--------------------------------------------------- %--------- if we have the input, we use it to determine amplitudes and %--------- frequencies if isempty(input) && isempty(amplitudes) utils.helper.err('You need to specify either an input signal, or a full description of the signals including amplitudes and frequecies.'); end computeFrequecies = false; if isempty(frequencies) computeFrequecies = true; end % Go through each ao for jj=1:numel(bs) if ~isa(bs.data, 'tsdata') utils.helper.err('quasiSweptSine requires time-series as input'); end [timespans, startTimes, stopTimes] = generateTimespans(bs(jj).t0, startTimes, stopTimes, durations); % Go through each time-span Txx = zeros(size(timespans)); dT = zeros(size(timespans)); for kk=1:numel(timespans) if isa(input, 'ao') inseg = input.split(plist('timespan', timespans(kk))); else % We don't have an input, so we need to create inputs from the % signal specs inseg = ao(plist('waveform', 'sine wave', ... 'nsecs', timespans(kk).interval, ... 'fs', bs(jj).fs, ... 'A', amplitudes(kk), ... 'f', frequencies(kk), ... 'toff', 0, ... 't0', startTimes(kk), ... 'yunits', inUnits)); end if computeFrequecies b = sineParams(inseg, plist('N', 1)); frequencies(kk) = b.y(2); end utils.helper.msg(msg.PROC1, 'Computing TF at %g Hz', frequencies(kk)); % Window w = ao(plist('win', win, 'length', inseg.len)); % Output outseg = bs(jj).split(plist('timespan', timespans(kk))); % Compute DFT fs = outseg.data.fs; N = outseg.len; J = -2*pi*1i.*(0:N-1)/fs; outxx = exp(frequencies(kk)*J)*(w.y.*outseg.data.getY); inxx = exp(frequencies(kk)*J)*(w.y.*inseg.data.getY); Txx(kk) = outxx./inxx; % Compute error % nout = noiseAroundLine(outseg, frequencies(kk), Nerror); nin = noiseAroundLine(inseg, frequencies(kk), Nerror); snr1 = abs(outxx)./nout; snr2 = abs(inxx)./nin; dT(kk) = abs(Txx(kk)) * sqrt( (1./snr1)^2 + (1./snr2)^2); end % End loop over timespans (segments) % Make output bs(jj).data = fsdata(frequencies, Txx, bs(jj).data.fs); bs(jj).data.dy = dT; bs(jj).data.setXunits('Hz'); bs(jj).data.setYunits(outseg.yunits./inseg.yunits); % Set name bs(jj).name = sprintf('sweptsine(%s,%s)', input.name, ao_invars{jj}); % Set procinfo bs(jj).procinfo = plist('frequencies', frequencies, ... 'timespan', timespans); % Add history bs(jj).addHistory(getInfo('None'), pl, ao_invars(jj), bs(jj).hist); end % Loop over input aos % set outputs varargout{1} = bs; end function n = noiseAroundLine(sig, f0, N) xx = abs(fft(sig)); % get the noise floor around the frequency of interest % - we need the bin that is nearest the frequency f = abs(xx.x - f0); [m, mi] = min(f); % xx.x(mi) % iplot(xx) % plot(xx.x(mi), xx.y(mi), 'x') M = N-1; % Measure below the line frequency if we can nl = 0; if (mi>1) ls = max(1, mi-2*M); le = max(1, mi-M); nl = xx.y(ls:le); end % Measure above the line frequency if we can nu = 0; if mi<xx.len us = min(xx.len, mi+M); ue = min(xx.len, mi+2*M); nu = xx.y(us:ue); end n = mean([nl;nu]); end function [ts, startTimes, stopTimes] = generateTimespans(t0, startTimes, stopTimes, durations) if isempty(durations) && numel(startTimes) ~= numel(stopTimes) utils.helper.err('You need to specify the same number of start and stop times.'); end if isempty(stopTimes) && numel(startTimes) ~= numel(durations) utils.helper.err('You need to specify the same number of durations and stop times.'); end ts = timespan.initObjectWithSize(1,numel(startTimes)); % Convert to time objects if iscell(startTimes) startTimes = time(startTimes); end if isnumeric(startTimes) newStarts = time.initObjectWithSize(1,numel(startTimes)); for kk=1:numel(startTimes) newStarts(kk) = t0+startTimes(kk); end startTimes = newStarts; end if isempty(durations) && iscell(stopTimes) stopTimes = time(stopTimes); end if isempty(durations) && isnumeric(stopTimes) if isnumeric(stopTimes) newStops = time.initObjectWithSize(1,numel(startTimes)); for kk=1:numel(stopTimes) newStops(kk) = t0+stopTimes(kk); end stopTimes = newStops; end end useDuration = isempty(stopTimes); for kk=1:numel(startTimes) if useDuration ts(kk) = timespan(startTimes(kk), startTimes(kk)+durations(kk)); else ts(kk) = timespan(startTimes(kk), stopTimes(kk)); end end end % get info object function ii = getInfo(varargin) if nargin == 1 && strcmpi(varargin{1}, 'None') sets = {}; pl = []; else sets = {'Default'}; pl = getDefaultPlist; end % build info object ii = minfo(mfilename, 'ao', 'ltpda', utils.const.categories.op, '$Id: quasiSweptSine.m,v 1.10 2011/04/08 08:56:16 hewitson Exp $', sets, pl); end % get default plist function plout = getDefaultPlist() persistent pl; if exist('pl', 'var')==0 || isempty(pl) pl = buildplist(); end plout = pl; end function pl = buildplist() % default plist for linear fitting pl = plist(); % Input p = param({'input', 'The input data series.'}, paramValue.EMPTY_DOUBLE); pl.append(p); % Start times p = param({'Start Times', 'A cell array of start times, or an array of time objects.'}, ... paramValue.EMPTY_CELL); pl.append(p); % Stop times p = param({'Stop Times', 'A cell array of stop times, or an array of time objects.'}, ... paramValue.EMPTY_CELL); pl.append(p); % Durations p = param({'Durations', 'An array of durations that can be used instead of the stop times.'}, ... paramValue.EMPTY_DOUBLE); pl.append(p); % Amplitudes p = param({'Amplitudes', 'An array of amplitudes.'}, ... paramValue.EMPTY_DOUBLE); pl.append(p); % Frequencies p = param({'Frequencies', 'An array of frequencies [Hz].'}, ... paramValue.EMPTY_DOUBLE); pl.append(p); % Input units p = param({'Input Units', 'If you don''t give an input signal AO, you can specify the units of the signal that will be constructed internally.'}, ... {1, {'V'}, paramValue.OPTIONAL}); pl.append(p); % Window p = param({'Win', 'A window to apply to each segment when computing the DFT.'}, ... paramValue.WINDOW); pl.append(p); % Error samples p = param({'Nerror', ['The number of samples either side of the line frequency to use to estimate the noise floor.'... 'The noise is estimated from <br><br><tt>mean([y(idx-2*M:idx-M);y(idx+M:idx+2M)])</tt><br><br> where <tt>M=N-1</tt> and <tt>idx</tt> is the index '... 'of the bin nearest to the frequency of the signal.']}, ... paramValue.DOUBLE_VALUE(5)); pl.append(p); % % Phases % p = param({'Phases', 'An array of phases [degrees].'}, paramValue.DOUBLE_VALUE(0)); % pl.append(p); end