Abstract :
[en] The present study exploits the maximum likelihood identification framework for deriving statistically-optimal models of nonlinear mechanical systems. The identification problem is formulated in the frequency domain, and model parameters are calculated by minimising a weighted least-squares cost function. Initial values of the model parameters are obtained by means of a nonlinear subspace algorithm. The complete identification methodology is first demonstrated on a Duffing oscillator, prior to being applied to a full-scale aerospace structure.
Scopus citations®
without self-citations
1