摘要
考虑两类双阈值二元神经网络模型,研究了具变号型初值解的渐近行为。结果表明:对大阈值情形,模型有唯一平衡点;对临界阈值情形,模型有两个或三个平衡点。进一步,在临界阈值情形,对时滞τ、阈值σ及初值Φ建立了保证系统趋于不同平衡点之一的充要条件。这些结论对相关文献作了较大推广。
Concentrating on the case where φ-σ and ψ-σ have sign changes on [-τ, 0],two types of models of neural networks with two thresholds and two neurons are considered. It's found that every solution (x^Ф(t) ,y^Ф(t) ) (Ф= (φ,ψ) is initial value) single equilibrium if the absolute values of threshold values are large. In particular, some necessary and sufficient conditions are established, in terms of r and Ф, which guarantee (x^Ф(t),y^Ф(t)) tends to those equilibria respectively in the critical case.
出处
《广西师范大学学报(自然科学版)》
CAS
北大核心
2006年第2期40-43,共4页
Journal of Guangxi Normal University:Natural Science Edition
基金
国家自然科学基金资助项目(10271044)
关键词
神经网络
时滞微分方程
渐近性
neural network
delayed differential equation
asymptotic behavior