Can System Identification Toolbox do transfer function estimation based on measured FRF, without input / output data?

I have measured the vibration frequency response function of a structure and exported the response function as shown below.

I want to approximate the result to a 2DOF vibration model. However, Model Estimating Using Frequency-Domain Data and tfest both require iddata as input, which has to include input and output signals. Is there a functionality for modal estimating which accepts measured frequency response as input?

[Mathieu NOE]

hello

as a worst case scenario, build yourself the 2 dof model and use fminsearch or other fitting tool to do the job

Answer 1

[Kartik Saxena]

Hi,

One approach I can think of, to approximate the result to a '2DOF' vibration model using the measured frequency response data, is to convert the frequency response data into a transfer function representation and then use the 'tfest' function to estimate the parameters of the '2DOF' model.

Here's an example of how you can do this:

% Convert frequency response data to complex form
resp = mag .* exp(1i * angle);

% Create an idfrd object from the frequency response data
frd_data = idfrd(resp, freq);

% Estimate the transfer function model using tfest
model = tfest(frd_data, 2);

In this example, we first convert the magnitude and phase response data into complex form. Then, we create an 'idfrd' object from the complex frequency response data. Finally, we use the 'tfest' function to estimate the transfer function model with 2 degrees of freedom.