Structure identification of block-oriented nonlinear systems using periodic test signals

Block-oriented nonlinear systems form an important class of nonlinear models, as they often reflect the physical structure of practical systems. The key problem when applying such models is to identify their structure. With the improvement of computational and graphical facilities it is possible to deduce information about the model structures from a graphical representation of their measured Volterra kernels. An introduction is given to the measurement of Volterra kernels using periodic multisine signals, as well as their evaluation and representation in the frequency domain. This is followed by an analysis of how certain structures relate to significant features in the graphical representation of second- and third-order kernels. An example is given for a simulated system