项目名称: 广义反应扩散神经网络的复杂动力学与同步控制
项目编号: No.61305076
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 甘勤涛
作者单位: 中国人民解放军军械工程学院
项目金额: 23万元
中文摘要: 作为一类具有广泛应用背景的非线性动力系统,神经网络的动力学行为是其应用和设计的基础。目前关于神经网络的研究大多集中于局域神经网络模型,而静态模型的研究相对较少,同时局域神经网络的研究结果也不适用于静态神经网络。为此,本项目拟借助非线性系统理论和控制理论等方法,结合数学模型分析和计算机模拟等手段从事以下研究:(1)建立能够包含静态神经网络和局域神经网络的广义反应扩散神经网络模型,揭示反应扩散效应对系统动力学的影响规律;(2)基于所建模型,针对时滞、参数不确定、随机扰动、Markov跳变以及脉冲作用等情况,提出一系列新的易于验证的稳定性判据;(3)研究广义反应扩散神经网络的混沌同步问题,确定影响同步的关键特征量,提出系统的同步控制策略并应用于保密通信。本项目的研究是目前神经网络动力学研究的新的探索,不仅可以推动神经网络理论的发展与完善,而且可以拓展神经网络的应用范围,具有重要的理论和应用价值。
中文关键词: 神经网络;稳定性;混沌;同步控制;反应扩散
英文摘要: The dynamics of neural networks, which has been applied widely, is the foundation of application and design. However, most of the researches have justly focused on the local field neural network model. Few researches have been done on studying the dynamical behaviors of static neural network model. Furthermore, the results obtained from the local field neural network model may be inapplicable to the static neural network model. Therefore, based on nonlinear system theory and control theory together with mathematical modeling and computer simulation, this project intends to: (1) establish the mathematical models of reaction-diffusion neural networks, which include two classes of fundamental neural networks, i.e., static neural networks and local field neural networks, as their special cases, further analyze the influence of reaction-diffusion terms on dynamical behaviors; (2) propose new and less conservative stability criteria for generalized reaction-diffusion neural networks with time delays or parameter uncertainties or stochastic perturbations or Markovian switching or impulse; (3) investigate the synchronization problems of generalized reaction-diffusion neural networks, design the communication security schemes based on chaos synchronization. This project is a new exploration of neural networks dynamics, w
英文关键词: neural networks;stability;chaos;synchronization control;reaction diffusion