Circuit Construction And Simulation Of Memristive Synapse Associative Memory Neural Network

The memristor has been proposed by professor Chua in 1971,because of the immature semiconductor,until 2008,the nano-scale memristor has been produce by HP laboratory.Due to its low power consumption,nonvolatile,nano scale,and nonlinear characteristic,it has promising prospect in chaotic circuit,secure communication,artificial neural network and storage device.memristor has many extremely similar characteristics with biological synapse,therefore,memristor can be used as artificial synapse in neural network for enhancing artificial intelligence system.Most of the memristive models are not perfect enough to simulate synaptic function.At the same time,the current structure of impulsive neurons is considerably complex,which is not conducive to large-scale integration.In this paper,an improved memristor model and a pulse neuron based on single electron transistor are proposed and applied to associative memory neural network.Furthermore,an ideal window function is proposed,which is applied to HP memristive model.Two neuron connection with memristor synapse are analyzed and extended to four neuron networks.The core work of this study is as follows:(1)A learning experience memristor(LEM)has been proposed for using as synapse in associative neural network.The properties of LEM are discussed under different external voltages.And then,we design a new feedback learning rule,All Input Feedback(AIF).An associative neural network based the AIF law and LEM synapse is constructed and analyzed.The properties of LEM are also verified through PSPICE simulation.(2)The associative neural network circuit based on AIF law and LEM are constructed and simulated using PSPICE,the simulation results are analyzed sufficiently.All simulation results show that the associative neural network incorporating LEM synapses and AIF learning law exhibits good performance,mimicking biological neural networks and self-learning behavior.(3)A spiking neuron circuit based on single electronic transistor(SET)and modified memristor synapses is proposed in this work.A novel window function is designed and used in our memristor model,which improves the linearity of the memristor and solves the boundary and terminal lock problems.In addition,we modify the memristor synapse for better weight controlling.Finally,to test the SNN constructed with the SET and memristor synapse,an associative memory learning process is implemented in the circuit based on PSPICE simulator.The associative memory activities were realized in the circuit simulation.These simulations not only provide an effective neural network circuit design,but also provide a way for the combination application of SETs and memristors.