An adaptive learning approach to the identification of structural and mechanical systems (Q1179377)

Considered are systems whose models are assumed to be known in structure. The problem consists in the identification of the values of the parameters describing the models. For this a set of response vectors \(\{r_ 1,r_ 2,\dots,r_ n\}\subset\mathbb{R}^ \nu\) and a set of parameter vectors \(\{p_ 1,p_ 2,\dots,p_ n\}\subset\mathbb{R}^ u\) is used to generate a linear, time invariant approximation in the form \(p=Mr\). In order to prevent the memory matrix \(M\) from being ill conditioned a multicriterion cost function is added to the problem. The paper makes use of notions known in the field of neural networks.