Title :
Practicable identification of nonstationary nonlinear systems
Author :
Marmarelis, V.Z.
Author_Institution :
University of Southern California, Graduate Center for Engineering Sciences, Department of Biomedical Engineering, Los Angeles, USA
fDate :
9/1/1981 12:00:00 AM
Abstract :
A method is presented that makes practicable the identification of nonstationary Wiener systems. The method employs a single-record estimator for the evaluation of the time-variant system kernels, thus relaxing the strict requirement of the ergodic hypothesis that has so far limited applications of the Wiener theory to stationary systems. The estimator is unbiased and consistent under fairly weak conditions for the system function, thus opening a host of new possibilities in the analysis of nonstationary phenomena.
Keywords :
nonlinear systems; parameter estimation; white noise; ergodic hypothesis; identification; nonlinear systems; nonstationary Wiener systems; single-record estimator; time-variant system kernels;
Journal_Title :
Control Theory and Applications, IEE Proceedings D
DOI :
10.1049/ip-d.1981.0045
