One possibility is the reverse finite element model based on an optimization procedure. You can use zero, first or second order optimization techniques (simplexes, neuron. networks, gen. alg., gradient, hessian methods). The main idea is to minimize the difference between your experimental output data and the numerical one respect to some model parameters (material, geometric parameters of the numerical model).
There are n number of methods to perform model updating of FRF data. These methods use modal data which are obtained by modal analysis of measured FRFs. The updating parameters corresponding to an analytical model are corrected to bring the analytical modal data closer to the experimentally derived one. For more details you can refer this paper.
I have done model updating based on natural frequency using Particle Swarm Optimization and Penalty function (Sensitivity based) method. Now, I want to do it for FRF data only, without using modal analysis.
You can use Response function method, proposed by Lin and Ewins. It uses measured FRF data directly without requiring any modal extraction to be performed.
Inverse of accurate dynamic stiffness matrix is FRF. Accurate dynamic stiffness matrix is combination of mass and stiffness matrices with square of frequency. Dynamic stiffness matrix is updated in each frequency.
Frequency response is the quantitative measure of the output spectrum of a system or device in response to a stimulus, and is used to characterize the dynamics of the system.
Estimating the frequency response for a physical system generally involves exciting the system with an input signal, measuring both input and output time histories, and comparing the two through a process such as the Fast Fourier Transform (FFT). One thing to keep in mind for the analysis is that the frequency content of the input signal must cover the frequency range of interest or the results will not be valid for the portion of the frequency range not covered.