up_freq, up_ifreq#
Purpose
Sensitivity and iterative updates based on a comparison of modal frequencies.
Syntax
[coef,mode,freq]=up_freq('Method',fID,modeID,sens); [coef,mode,freq]=up_ifreq('Method',fID,modeID,sens);
Description
up_freq and up_ifreq seek the values coef of the currently declared Up parameters (see the upcom Par command) such that the difference between the measured fID and model normal mode frequencies are minimized.
Currently 'basic' is the only Method implemented. It uses the maximum MAC (see ii_mac) to match test and analysis modes. To allow the MAC comparison modeshapes. You are expected to provide test modeshapes modeID and a sensor configuration matrix (initialized with fe_sens).
The cost used in both functions is given by
norm(new_freq(fDes(:,1))-fDes(:,2))/ norm(fDes(:,2))
up_freq uses frequency sensitivities to determine large steps. As many iterations as alternate matrices are performed. This acknowledges that the problem is really non-linear and also allows a treatment of cases with active constraints on the coefficients (minimum and maximum values for the coefficients are given in the upcom Par command).
up_ifreq uses any available optimization algorithm (see upcom opt) to minimize the cost. The approach is much slower (in particular it should always be used with a reduced model). Depending on the algorithm, the optimum found may or may not be within the constraints set in the range given in the upcom Par command.
These algorithms are very simple and should be taken as examples rather than truly working solutions. Better solutions are currently only provided through consulting services (ask for details at info@sdtools.com).
See also