Hindmarsh-Rose神经元网络斑图动力学研究

发布时间：2018-11-26 07:22

【摘要】：神经系统是由神经元组成的,神经元之间通过各种耦合方式来实现信号传递。由于靶波和螺旋波可以像“节拍器”一样,调节神经系统电活动的集体行为,因此斑图动力学理论能够预测网络群体振荡行为的稳定性。当正常信号传播受到干扰或出现非自发的神经疾病时,网络中呈现混沌状态。本文基于Hindmarsh-Rose神经元网络模型构造了最近邻连接下的规则神经元网络,通过利用自适应控制、调节系统参数、改变边界条件以及实验条件等方法实现对神经元网络稳定性控制。针对该课题做了以下研究:1.通过三种不同的方法产生靶波,研究靶波在易激发介质中出现的潜在机制。并加入与结构有关的人工缺陷观察对靶波的影响。证实了缺陷对靶波的影响取决于靶波的内禀属性(产生靶波的方式)。2.神经系统缺陷可以发射连续波或脉冲,扰乱正常神经系统的信号传播。通过对外部激励产生波及其传播过程的研究,分析缺陷的形成机制。发现在外部激励下可以产生缺陷,且外部激励产生的波与缺陷诱发的波可以共存。3.网络边界上的结点选取随机初始值,通过观察神经元网络各结点膜电位的空间分布,来研究初始值对神经元放电模式的影响。发现适当的耦合强度下,网络中可以观察到螺旋波。4.研究规则神经元网络中同时施加局部周期激励和噪声时的动力学行为。发现螺旋波和平面波可以共存于神经元网络中,即使施加的噪声和外界周期激励位置不同,二维规则神经元网络中仍会出现随机共振行为。5.对网络少数结点的输出信号进行动态实时检测,利用非线性分析方法诊断网络细微涨落引发的相变行为,以及通过各个采样检测点的关联度,确定故障和崩溃的位置和范围。把该方法扩展到多体系统安全性检测和诊断,为消除故障、减少事故损害提供依据。
[Abstract]:The nervous system is composed of neurons, which transmit signals by various coupling methods. Because target wave and spiral wave can adjust the collective behavior of nervous system electrical activity just like metronome, the theory of pattern dynamics can predict the stability of population oscillation behavior in the network. When the normal signal propagation is disturbed or the non-spontaneous neural disease occurs, the network presents chaotic state. Based on the Hindmarsh-Rose neural network model, a regular neural network with nearest neighbor connection is constructed in this paper. By using adaptive control, the parameters of the system are adjusted. The stability control of neural network is realized by changing the boundary condition and experimental conditions. The following research has been done on this subject: 1. The potential mechanism of target wave in excitable medium is studied by three different methods. The effect of artificial defect observation on the target wave is added. It is proved that the effect of defect on target wave depends on the intrinsic property of target wave (the way in which the target wave is produced). 2. Nervous system defects can emit continuous waves or pulses to disrupt signal transmission in the normal nervous system. The mechanism of defect formation is analyzed by studying the wave generation and its propagation process of external excitation. It is found that defects can be produced under external excitation and that waves generated by external excitation and waves induced by defects can coexist. The random initial values of the nodes in the network boundary are selected to study the effect of the initial values on the firing patterns of neurons by observing the spatial distribution of the membrane potentials of each node of the neural network. It is found that helical waves. 4. 4 can be observed in the network with proper coupling strength. The dynamic behavior of regular neural networks under local periodic excitation and noise is studied. It is found that helical wave and plane wave can coexist in the neural network, even if the noise applied is different from the external periodic excitation position, the stochastic resonance behavior will still occur in the two-dimensional regular neural network. The output signals of a few nodes in the network are dynamically detected in real time. The nonlinear analysis method is used to diagnose the phase change behavior caused by the subtle fluctuations of the network. The location and scope of the fault and collapse are determined by the correlation degree of each sampling and detection point. The method is extended to multibody system safety detection and diagnosis, which provides the basis for eliminating fault and reducing accident damage.
【学位授予单位】：兰州理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R338;TN911.6

【参考文献】