id_dspi

Contents Functions

id_dspi

Purpose

Direct structural system parameter identification.

Syntax

[a,b,c,d] = id_dspi(y,u,w,idopt,np)

Description

The direct structural system parameter identification algorithm [54] considered here, uses the displacement frequency responses y(s) at the different sensors corresponding to the frequency domain input forces u(s) (both given in the xf format). For example in a SIMO system with a white noise input, the input is a column of ones u=ones(size(w)) and the output is equal to the transfer functions y=xf. The results of this identification algorithm are given as a state-space model of the form

{

ṗ

} = [

0	I
−K_T	−C_T

] {

ṗ

} + [

b_T

] {u} and {y} = [

c_T

] {

ṗ

}

where the pseudo-stiffness K_T and damping C_T matrices are of dimensions np by np (number of normal modes). The algorithm, only works for cases where np is smaller than the number of sensors (ci.IDopt.ns).

ci=iicom('curveload sdt_id');
R1=ci.Stack{'Test'};
[a,b,c,d] = id_dspi(R1.xf,ones(size(R1.w)),R1.w,R1.idopt,4);

For SIMO tests, normal mode shapes can then be obtained using [mode,freq] = eig(-a(np+[1:np],1:np)) where it must be noted that the modes are not mass normalized as assumed in the rest of the Toolbox and thus cannot be used directly for predictions (with nor2xf for example). Proper solutions to this and other difficulties linked to the use of this algorithm (which is provided here mostly for reference) are not addressed, as the main methodology of this Toolbox (id_rc, id_rm, and id_nor) was found to be more accurate.

For MIMO tests, id_dspi calls id_rm to build a MIMO model.

The identification is performed using data within ci.IDopt.SelectedRange. y is supposed to be a displacement. If ci.IDopt.DataType gives y as a velocity or acceleration, the response is integrated to displacement as a first step.