忆阻Hopfield神经网络的混沌特性研究

Study on chaotic characteristics of memristive Hopfield neural network

Hopfield神经网络能很好地模拟大脑中的混沌行为,本文根据突触间存在的串扰行为,设计了一种四元忆阻Hopfield神经网络。运用Lyapunov指数谱和分岔图等方法分析忆阻参数和串扰参数变化对神经网络动力学特性的影响,发现该神经网络出现一系列复杂混沌现象。Pspice电路仿真实验结果与理论分析一致,进一步揭示了忆阻Hopfield神经网络的可实现性。

The evidence shows that there exist complex chaos dynamics behaviors in the brain,and neural networks working under the chaotic mechanism can enhance the complexity of computation.Hopfield neural network is an important model of artificial neural computation,and it can excellently simulate the chaotic behavior in the brain.The memristor has become one of the best devices for simulating neuronal synapses not only because its resistance can be adjusted,but also because it has memory characteristics.A four-element Hopfield neural network based on memristor is proposed to make the neural network closer to the practical biology.By means of Lyapunov exponential spectrum and bifurcation diagram,the influence of system parameters on the dynamic characteristics is analyzed in detail,and it is found that the neural network model produces a series of complex chaotic phenomena.The Pspice simulation results are consistent with the results of theoretical analysis,which further reveals the realizability of the memristive Hopfield neural network,and leads to a good prospect in simulating brain neural activity and memory behavior.