Abstract: The?exponential?stability?of?delayed?bidirectional?associative?memory?neural?networks?（?DBAM?neural?networks）?is?studied?and?the?influence?of?pulse?on?exponential?stability?is?focused.?The?result?shows?that?if?the?whole?exponent?of?non?-?pulse?DBAM?is?stable,?the?corresponding?pulse?system?can?remain?its?stability,?even?if?the?state?are?magnified?to?a?larger?scale?at?the?pulse?instant.?Furthermore,?the?pulse?upper?bound?and?pulse?interval?which?can?keep?the?exponents?of?the?system?stable?is?estimated,?according?to?the?degree?of?exponential?convergence?of?non-pulse?system.

Key words: delayed bidirectional associative memory neural networks （DBAM neural networks） , exponential stability , time-delay , pulse stability ;

摘要： 研究了时滞双向联想神经网络的指数稳定性,集中讨论了脉冲对指数稳定性的影响。结果显示,若非脉冲 DBAM系统全局指数稳定,即使状态在脉冲时刻被放大到某一较大范围,相应的脉冲系统也能保持其稳定特性。根 据非脉冲系统的指数收敛度,对保持系统指数稳定的脉冲上界和脉冲间隔进行了估计。

关键词: 双向联想神经网络(BAM神经网络), 指数稳定性, 时滞, 脉冲稳定性