连续非对称神经网络的动力学特性
DYNAMIC PROPERTIES OF CONTINUOUS UNSYMMETIC NEURAL NETWORKS
-
摘要: 本文讨论了连续非对称神经网络的动力学特性,给出了神经网络有唯一平衡点的条件,并讨论了当连接距阵T变化时不产生静态分叉和Hopf分叉的条件,给出了平衡点的全局渐近稳定性和全局指数稳定性的充分条件。Abstract: The dynamic properties of continuous unsymmetic neural networks are discussed, and the condition of existence of unique equilibrium point is obtained. The conditions not producing static state bifurcation and Hopf bifurcation are put forward and the sufficient conditions of overall asymptotic stability and exponential stability are obtained.
-
Li J H, Micbel A N, Porod W. IEEE Trans, on CAS, 1988, CAS-35(8); 976-986.[2]Matuoka K. Neural Networks, 1992, 5(3): 995-500.[3]张本想,肖国镇.西安电子科技大学学报,1992,19(1): 28-34.[4]蒋耀林,徐建学.非对称连续时间神经网络系统的稳定性分析.中国神经网络1993年学术大会论文集,西安: 228-232.[5]陆启韶.常微分方程的定性方法和分叉.北京:北京航空航天大学出版社,1989,231-238,277-278.[6]焦李成.神经网络系统理论.西安:西安电子科技大学出版社,1990, 59-67.
计量
- 文章访问数: 1922
- HTML全文浏览量: 105
- PDF下载量: 395
- 被引次数: 0