A Behavioral Framework for Data-Driven Modeling of Nonlinear Systems in Vector-Valued Reproducing Kernel Hilbert Spaces
It provides a theoretical framework for data-driven modeling of a broad class of nonlinear systems, but the contribution is primarily theoretical and incremental.
The paper generalizes Willems' behavioral approach to nonlinear systems in vector-valued RKHS, covering Volterra series and Hammerstein models, and links it to data-driven methods like minimum-norm interpolation and subspace identification for modeling without explicit system identification.
We generalize Jan Willems' behavioral approach to a class of discrete-time nonlinear systems in a vector-valued reproducing kernel Hilbert space (RKHS). Apart from linear time-invariant systems, this class covers nonlinear systems modeled by Volterra series and their autoregressive variants, as well as systems admitting Hammerstein-type state-space realizations. We apply the proposed framework to the problem of data-driven modeling of such systems, i.e., when simulation or control objectives for an unknown system are carried out without an explicit system identification step. To that end, we link the behavioral approach to two data-driven modeling methods in a vector-valued RKHS: (1) minimum-norm interpolation and (2) subspace identification.