Design Of A Synaptic Circuit Based On Memristor In Pulse Neural Networks

In the past ten years,many researchers have conducted the researches about the modeling,applications and simulation of the memristor.There are lots of domestic researches on the memristor currently.These researches have made a great achievement.At present,the kind of the models of the memristor to be studied is simple.The HP memristor model which is a t ype of non-linear models of the memristor.Domestic researches on other types of the memristor such as the manufacturing and simulation are far from those abroad.In this paper,we build and analyze the mathematical models and circuits related to the memristor.Also,we do the research about the circuit characteristics.Based on the model of the memristor,the coupling structures of the memristor and the coupling circuits' characteristics are studied.On the basis of circuits' characteristics,the design of the synapse circuit for the pulsed neural network in a variety of environments can be done.In this paper,the models of the memristor were studied,including the HP model and the Exponential memristor.Based on the matlab software platform,the characteristics of the memristor itself and the influence of changes in external parameters were studied and simulated.The characteristics of memristors on which they are subjected to different excitations are mainly studied.Under the influence of pulse excitatio n,the boundary effect of memristor is studied.The model of the memristor is modified through the window functions.The differential corrections of the memristor are implemented by choosing the different window functions.The memristor's analog circuit is built in the PSPICE environment and the simulation analyses are performed.Based on the controlled source,there is a research on the coupling circuits with different types.The coupling characteristics of the memristor and the application of the circuits were studied.The bridge circuit of the memristor is improved.The STDP rule which is based on the m achine learning is studied.The memristor is studied with the STDP rule.Circuits are designed by studying the forms and states of the coupling circuits and the synaptic circuits of the memristor.The bridge circuit and the anti-series circuit are designed.The paper studies the dynamic characteristics of the synapse circuit and its advantages and disadvantages.The STDP characteristics of the synaptic circuit applied in the neural network are studied and analyzed.A memristor-based circuit structure for synapses in neural networks is proposed and designed to accommodate more application scenarios.The studies of circuit design and application are inseparable from the analyses of the circuit's performance and structure.The synaptic circuit designed in this paper has the higher degree of freedom and the more flexible mode of work,meanwhile,it has the more complex architecture and the higher requirements of resource.