Stochastic Approximation in Hilbert Space: Identification and Optimization of Linear Continuous Parameter Systems

From MaRDI portal





The paper deals with a system given by the equation \[ J_ n=\int^{T}_{0}K(\tau)z(n\Delta -\tau)d\tau +\psi_ n \] where z(\(\cdot)\) denotes the input, K(\(\cdot)\) the transfer function, \(\Delta\) the sampling interval, \(y_ n\) the output (at the sampling time \(n\Delta)\), \(\{\psi_ n\}^ a \)stationary (observation noise) sequence with mean zero and independent of z(\(\cdot)\). It is assumed that the transfer function K(\(\tau)\) is unknown. An algorithm for recursive estimation of K(\(\tau)\) is presented under the former assumptions. This algorithm is analogous to a similar one in the finite dimensional case, of course. Further, convergence, rate of convergence and behaviour of the estimates for large times are studied in detail.











This page was built for publication: Stochastic Approximation in Hilbert Space: Identification and Optimization of Linear Continuous Parameter Systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3705329)