神经网络的混沌运动与控制

本文采用一种由混沌神经元构成的联想记忆神经网络．以混沌神经网络为基础，研究其非线 性动力学特性、混沌吸引子轨迹以及对初始条件的敏感性, 实现混沌神经网络的动态联想记 忆功能．在网络输入发生较大变异情况下所发生的失忆，本文采用时空系统混沌控制的钉扎 反馈方法，使网络恢复记忆．上述研究通过对异步电动机故障的动态记忆和恢复控制的仿真 实验得到证实．本文研究结果表明，在国内外对神经网络混沌控制研究的热点中，时空系统 的钉扎反馈控制是一种值得推荐的方法；神经网络的混沌控制扩大了网络的容错性，进而提 高了混沌神经网络的实用性，这将在复杂模式识别，图象处理等工程上具有广阔的应用前景 ．

CHAOS AND ITS CONTROL IN NEURAL NETWORK

An associative neural network is used with chaotic neural models interconnected through a conventional auto associative matrix of synaptic weights. Dynamic associative memory and essential characteristics of chaotic neural network is dealt with: nonperiodic chaos, chaotic attractors, and sensitivity to starting condition. And a spatiotemporal system is modeled by a neural network. Feedback pinning are used to control chaos of the system by studying a certain reference sample, that is beneficial to retrieving dynamic memory. In this paper, dynamic associative memory and retrieval for faults of three phase asynchronous motor with broken bars is simulated by chaotic neural network and its control.