Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

Identification of extended Hammerstein systems using dynamic self-optimizing neural networks.

Xuemei Ren¹, Xiaohua Lv

¹School of Automation, Beijing Institute of Technology, Beijing, China. xmren@bit.edu.cn

IEEE Transactions on Neural Networks

|June 29, 2011

Summary

A new dynamic self-optimizing neural network (DSONN) effectively models complex systems with non-Gaussian noise. This approach automatically adjusts network structure and weights for accurate system identification without prior knowledge.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Composite nonlinear disturbance observer-based practical prescribed time control for nonlinear systems with matched and mismatched disturbances.

ISA transactions·2026

Same author

Three-dimensional resolution enhancement of two-photon microscopy by combining point spread function engineering and multi-image deconvolution.

Optics letters·2026

Same author

Electrocatalytic Properties of Binary PdSn Nanoparticles for Ethanol Oxidation.

Langmuir : the ACS journal of surfaces and colloids·2026

Same author

Fruquintinib plus sintilimab in advanced cervical cancer: an open-label, multicentre, phase 2 study.

Communications medicine·2026

Same author

Seed Size-Number Trade-Off in Populations of a Cosmopolitan Species (<i>Oxybasis glauca</i>) From Xinjiang, China.

Ecology and evolution·2026

Same author

Surgical management of spinal metastases from primary lung carcinoma: demographics, clinical characteristics, and outcomes-A retrospective analysis.

Frontiers in surgery·2026

Area of Science:

Artificial Intelligence
Machine Learning
Control Systems Engineering

Background:

Extended Hammerstein systems often exhibit non-Gaussian noise, posing challenges for traditional modeling techniques.
Accurate system identification requires adaptive models capable of handling unknown nonlinearities and noise characteristics.

Purpose of the Study:

To propose a novel dynamic self-optimizing neural network (DSONN) for modeling extended Hammerstein systems with non-Gaussian noises.
To develop an online learning algorithm that simultaneously optimizes network structure and weights.
To achieve robust system identification without requiring a priori knowledge of the system's nonlinear dynamics.

Main Methods:

System order estimation to determine the input vector for the neural network.

Related Experiment Videos

Online generation of the hidden layer with adaptive growth and pruning steps for minimal realization.

Simultaneous adjustment of network weights and structure using weight variations as optimization criteria.

An integrated performance function combining identification error and entropy penalty for noise attenuation.

Main Results:

The proposed DSONN successfully models extended Hammerstein systems with non-Gaussian noises.
The algorithm demonstrates effective online adjustment of both network structure and weights.
Guaranteed convergence of weights is achieved through appropriate learning rate selection.
The method's efficiency is validated through applications to three distinct Hammerstein systems.

Conclusions:

The DSONN provides a powerful and flexible framework for adaptive system identification in the presence of complex noise.
The developed approach eliminates the need for prior knowledge of system nonlinearities, enhancing its practical applicability.
The DSONN offers a robust solution for modeling challenging dynamic systems.