Researsh On Disturbance Rejection Of Small-world Spiking Neural Network Based On Synaptic Plasticity

With the increasing complexity of electromagnetic environment,the shortcomings of traditional anti-electromagnetic interference methods for electronic systems are becoming increasingly prominent.The organism under the regulation of neural system has advantages on self-adaptive mechanism and disturbance rejection ability.Based on these good characteristics of organism,constructing a neural network with dynamic regulation mechanism of synaptic plasticity to study its dynamic regulation law and anti-interference function provides a new method for improving the stability of electronic systems in complex electromagnetic environment.In this study,the spiking neural network with the regulation of both excitatory synaptic plasticity and inhibitory synaptic plasticity is constructed.The dynamic evolution of the network under the regulation of synaptic plasticity is analyzed.Firing rate and correlation between the membrane potential are considered as disturbance rejection indexes to evaluate disturbance rejection ability under gauss white noise stimulation,square wave impulse stimulation,AC electric field stimulation and neuron damage.The main work of this study is as follows:1.Izhikevich neurons are used as node models,excitatory synaptic plasticity and inhibitory synaptic plasticity are used as the connection relationship,based on the small-world topology the small-world spiking neural network with dynamic regulation is constructed.The evolution process of the dynamic characteristics of the small-world spiking neural network under different simulation time is analyzed.The experimental results show that the network characteristic parameters change dramatically during the initial period of simulation.With the increase of simulation time,the network characteristic parameters change gradually smooth and eventually tend to be stable.2.The disturbance rejection ability of the small-world spiking neural network under gauss white noise stimulation,square wave pulse stimulation and AC electric field stimulation is analyzed from two aspects of firing rate and correlation between the membrane potential.The experimental results show that,based on the dynamic regulation of synaptic plasticity,the external stimulus within a certain range has little impact on the network,when the external stimulus exceeds this range,it has a greater impact on the network,small-world spiking neural network has disturbance rejection ability under external stimulus within a certain range.The dynamic regulation process of synaptic plasticity weights is significantly correlated with the disturbance rejection ability of the neural network.3.The disturbance rejection ability of the small-world spiking neural network under random attack and deliberate attack is analyzed from two aspects of firing rate and correlation between the membrane potential.The experimental results show that,under the random attack,a certain number of neuron damage has little impact on the network,when the number of neuron damage exceeds this range,it has a greater impact on the network,small-world spiking neural network has disturbance rejection ability under neuron damage within a certain number.Under the deliberate attack,the node with low degree of damage has less impact on the neural network,the node with high degree of damage has greater impact on the neural network.