.png){kind=icon}
Return
Rapid dynamical pattern modeling for sampled-data nonlinear systems using regression filter
.png){kind=icon}
PRE
AI
DOI：10.1007/s11071-024-10654-3.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper investigates a rapid dynamical pattern modeling method for a class of nonlinear sampled-data systems. Firstly, within the nonlinear sampled-data systems framework, the consistency condition is presented based on the approximate discrete-time model. Then, a regression filter-based dynamic learning method is proposed to enhance online learning performance of neural networks (NNs). The exponential convergence of NN estimated weights, which stem from regression filter-based weight update law, is deduced based on two newly derived corollaries. The unknown system nonlinear dynamics are accurately modeled using the dynamic learning method. The modeling knowledge is defined as training patterns which are expressed in a group of constant NNs. Analytical results concerning the pattern recognition condition and recognition time are derived based on the duty ratio, consistency condition, and feature interval. The stored knowledge can be reused for dynamical pattern recognition. The effectiveness of the proposed scheme is illustrated through the simulation results.
Keywords:
Sampled-data systems
Neural networks
Persistent excitation
Regression filter
Pattern recognition
Journal
IF:
6
Papers:
1.4W
Citations:
4.1W
Organization
Cited Papers
ShareSaveShareSave
ShareSaveData-driven learning of chaotic dynamical systems using Discrete-Temporal Sobolev Networks.png)
NEURAL NETWORKS
NEURAL NETWORKSIF6.3
PREAI
ShareSaveShareSaveShareSaveShareSave
View more