Adaptive nonlinear dynamical processing for time series analysis
Jeffrey S. Brush, James B. Kadtke · AIP conference proceedings · 1994
Adaptive, or time‐varying, modeling approaches to signal processing have typically been employed when non‐stationarity is presumed to exist. In a nonlinear dynamics framework, the adaptive paradigm has an additional application in modeling which simultaneously addresses some of the limitations of local linear and static global methods, even for stationary situations. These new methods therefore have the ability to account for non‐stationarity as well as nonlinear signal properties. We discuss the implementation of two continuous model update schemes, as well as applications to system characterization, parameter tracking, and transient detection in noise.