基于多态忆阻的新型前馈神经网络仿真电路

A Current-Type Feedforward Neural Network Circuit Based on Multi-Resistance Memristor Devices

基于多阻态忆阻器,提出了一种神经网络电路仿真实现方法。该方法采用理想忆阻器和高性能MOS管模型构建1T1M(1 Transistor 1 Memristor)的权重结构单元,结合低功耗运放和寄存器技术,设计了一种神经网络运算电路。该神经网络电路的层与层之间可以传递正负浮点数信号,实现对应的权值加和、偏置信号设置、分类以及信号激活功能。仿真实验同上位机对比的结果表明,该模型通过高速信号处理能力提升前向人工神经网络的运行性能。此外,在实现相同神经网络结构的情况下,不仅一定程度上避免了忆阻阵列中存在的漏电流现象,还有效节省了硬件资源。

Based on multi-resistive memristor,a method of neural network circuit simulation is proposed.Ideal memristor and high performance MOS tube models are used to construct the weight structure unit of 1T1M(1 Transistor 1 Memristor).Combined with low power operational amplifier and register technology,the analog circuit of neural network operation is designed.The positive and negative floating-point signals can be transmitted between layers of the neural network,and the functions of weights addition,bias signal setting,classification and signal activation can be realized.The simulation results show that the model can improve the performance of the forward artificial neural network through high speed signal processing.In addition,in the case of the same neural network structure,not only the leakage current phenomenon is eliminated partly,but also the hardware resources are saved.