Dynamic Behaviors Of Memristive Hopfield Neural Network Under The Effects Of Magnetic Fields And Synapses

As one of the neural networks closest to human brain,the memristive neural network has complex chaotic dynamic behaviors like brain,which plays an important role in understanding the working mechanism of brain.However,most of the existing memristive neural network models only consider a single biological function,and do not fully consider the complexity of the environment where neurons live.With the wide use of electronic equipment,the influence of magnetic field on artificial neurons is deepening day by day,and synapses regulate the behavior of neurons.They are all non-negligible factors in the construction of memristive neural network model.To this end,we propose three different memristive neural network systems,taking into account the effects of synapses and magnetic fields.The main contents of the paper are as follows:Firstly,considering that electromagnetic radiation is unavoidable in modern society and autosynapse is an effective means for neurons to regulate behavior,a simple memristive neural network is designed by introducing the effects of electromagnetic radiation and memristive synapses in Hopfield neural network.The system is characterized by the fact that one neuron is disturbed by electromagnetic radiation and the autosynapse of the other neuron is replaced by a memristor.We find that the system exhibits different periodic and chaotic states depending on the coupling parameters.At the same time,coexisting attractors and transient chaos dependent on initial condition exist in the system.We verify the correctness of the numerical simulation by building circuits.Secondly,considering the influence of external current stimulation on the neural network,a more complex memristive neural network is proposed based on the above system,and the used memristors are improved.The results show that the new system can show the coexistence of periodic attractors,quasi-periodic attractors and chaotic attractors.In particular,the system can generate an infinite number of different kinds of attractors with a hidden extreme multistability phenomenon by selecting certain parameters.In addition,the neural network has been implemented on Pspice and DSP boards.Finally,a two-neuron Hopfield neural network is proposed,considering the effects of electromagnetic induction and synaptic crosstalk.Electromagnetic induction is mainly generated by the voltage difference between two neurons,and its direction of action is bidirectional,which is more complex than electromagnetic radiation phenomenon.Synaptic crosstalk is caused by the synaptic interaction between two neurons,which is more complex than autosynaptic phenomenon.Through the nonlinear analysis methods,it is found that the system can produce multiple coexistence behaviors depending on the initial condition.Multiple chaotic attractors were found by adjusting system parameters.The correctness of the numerical simulation is verified by Pspice and DSP experiments.In conclusion,three memristive neural networks based on complex biological mechanisms are proposed.Dynamic studies show that these systems exhibit complex dynamic behaviors,such as coexisting attractors,extreme multistability and transient chaos.This not only provides a new way to explain brain neuroscience,but also brings more flexibility to the design of artificial neural networks,which is of great significance to promote the development of nonlinear dynamics.