一种阈值转换忆阻器仿真电路设计

Design of a threshold conversion memristor simulation circuit

人工神经元和突触是处理神经形态系统中复杂信息的关键单位。忆阻器经常被设计为人工突触,因为它们具有简单的结构,逐渐变换的电导和高密度的集成度。文章设计的这种阈值转换忆阻器(TSM)模型的仿真电路不仅可以尽可能高效的仿真出TSM的基本性能,并通过调节模型相关设置值来计算得出所需要模拟的器件对应参数值,同时解决了现有的仿真模型在仿真过程中仿真时间长、产生仿真文件大、设计难度大、应用的元器件多等技术难题。该模型仅包括电阻、小型运算放大器、滞回比较器、压控开关以及电源,其使用的元器件相对于传统设计电路大幅度减少,模型仿真时间以及仿真文件大小大幅度降低。同时本模型能够较为真实地拟合出TSM相应的电气特征,可以通过调节参数改变TSM的电阻值大小,从而在不同的外界工作条件下,实现记忆或者信息存储的功能。在应用于整合-激发神经元电路(IF)中,也可以实现预期所需要的模拟生物突触间的脉冲信号特性。这些结果表明,良好的人工突触电路可以实现生物突触的基本形态描述,并表现出神经形态计算的巨大潜力。

Artificial neurons and synapses are key units for processing complex information in neuromorphic systems. Memristors are often designed as artificial synapses because of their simple structure, gradually changing conductance, and high integration. In this paper, the design of this kind of Threshold Switching Memristor(TSM) model in the simulation circuit can not only efficiently simulate the basic performance of the TSM as far as possible, but also easily calculate parameters value of the device by adjusting the relevant Settings. At the same time, it can solve flaws of the existing simulation model during the simulation process, such as long-time simulation, large sized producing simulation files,difficulties of the designing,the large number of components and other technical problems.The model only includes resistors, small operational amplifiers, a hysteresis comparator, a voltage control switch and a power supply. Compared with the traditional design circuit, the components used in this model as well as the simulation time and the size of simulation files are greatly reduced. Meanwhile, the model can fit the corresponding electrical characteristics of the TSM more realistically, and the resistance value of the TSM can be changed by adjusting the parameters, so as to realize the function of memory or information storage under different external working conditions. The desired characteristics of synaptic impulses can also be achieved in the integration-and-fire(IF) neuron circuit. These results indicate that a good artificial synaptic circuit can realize the basic morphological description of biological synapses.It shows great potential in neuromorphic computation.