Title :
Persistency of excitation of identification with N-D sampling neural networks
Author :
Dzielinski, A. ; Zbikowski, R. ; Kaczorek, T.
Author_Institution :
Dept. of Mech. Eng., Glasgow Univ., UK
Abstract :
In this paper, persistency of excitation of a novel method of discrete-time NARMA control models identification is presented. It is proved that this can be restated as an invertibility problem for a Vandermonde type transformation matrix. A necessary and sufficient condition for the matrix invertibility is given. Consequences for neural network modelling of nonlinear control systems are highlighted
Keywords :
control system analysis; discrete time systems; matrix algebra; neural nets; nonlinear control systems; state estimation; N-D sampling neural networks; Vandermonde type transformation matrix; control simulation; discrete-time NARMA control models identification; excitation persistency; invertibility problem; neural network modelling; nonlinear control systems; Fourier transforms; Neural networks; Nonuniform sampling; Sampling methods; Shape;
Conference_Titel :
Industrial Electronics, 1996. ISIE '96., Proceedings of the IEEE International Symposium on
Conference_Location :
Warsaw
Print_ISBN :
0-7803-3334-9
DOI :
10.1109/ISIE.1996.548403
