摘要
提出了振荡型混沌神经网络的结构模型和动力方程 ,证明了该网络的稳定性和 Hopf分支的必要条件 ,得到了在三维模式空间中的轨迹 ,利用功率谱密度分析了重迭函数的波形 .试验结果表明输入模式在靠近和远离记忆模式两种情况时 ,网络具有不同的信息处理能力 .
The construction model and dynamics equation of oscillatory-chaotic neural network are proposed. The stability and the necessary condition of this network are proved. Orbits in the three dimensional pattern space are obtained. The waveforms of the overlap function are analyzed by using power spectrum density. It has been shown that the network possesses different information processing capacities when an input pattern is close to or far from a memory pattern. This network has very important application value in dynamics information processing of spatiotemporal patterns and intelligent information processing systems
出处
《自动化学报》
EI
CSCD
北大核心
2002年第3期401-407,共7页
Acta Automatica Sinica
基金
国家自然科学基金 ( 6 0 0 75 0 0 8
6 0 1 0 2 0 1 0 )资助
关键词
振荡型混沌神经网络
智能信息处理
混沌轨迹
极限环
Oscillatory-chaotic neural networks, intelligent information processing,chaotic orbit, limit cycle
