自旋忆阻CPSO-PID神经网络

The Spintronic Memristor-Based CPSO-PID Neural Network

采用混沌粒子群优化算法(CPSO)对网络初始值进行迭代寻优,加快系统的收敛速度.为了实现PID神经网络的硬件平台搭建,新型纳米器件——自旋忆阻器,由于其独特的记忆特性和纳米级尺寸,被用来模拟电子突触,以实现神经元之间的相互连接.最后,通过对强耦合系统进行解耦控制验证了该方案的有效性.

PID neural network combines the advantages of conventional PID controller and neural network,which can greatly control the multivariable time-varying systems.In this article,CPSO is used to initial the weights by iteration optimization for the network,accelerating the convergence speed of the system.For building the circuit of PID neural network,the novel nanoscale electronic device,Spintronic Memristor,due to its unique memory properties and nanoscale size,is used to perform the function of electronic synaptic,achieving the interconnection between neurons.Finally,the effectiveness of the proposed scheme in the paper is verified by decoupling the strong coupling system.