基于CMAC小脑神经网络的超磁致伸缩作动器高精度控制的仿真研究

High-precision control of giant magnetostrictive actuator based on CMAC neural network

为了补偿超磁致伸缩作动器(GMA)内在的滞回非线性提高其精度,将小脑神经网络(CMAC)前馈和PID反馈控制器相结合,提出了一种实时滞回补偿控制策略,以期实现GMA的高精度跟踪控制。由于CMAC神经网络不能够直接逼近滞回逆这种具有记忆性的多映射现象,通过引入一个滞回逆算子,将多映射的滞回逆转换成一一映射,然后运用CMAC神经网络控制器来逼近这个一一映射,从而建立一个基于CMAC神经网络的滞回逆模型。仿真结果表明该控制策略能适应GMA滞回特性随输入信号的变化,在线建立GMA的滞回逆模型,从而消除滞回非线性的影响,实现GMA的高精度控制。

In order to compensate its inherent hysteresis nonlinearity and improve its precision of a giant magnetostrictive actuator(GMA),a real-time hysteretic compensation control strategy was proposed,combining a feedforward cerebellar model articulation controller(CMAC) and a proportional integral derivative(PID) feedback controller to realize the precision position tracking control of the GMA.As CMAC neural network could not be used to approximate the multi-valued mapping of an inverse hysteresis directly,an inverse hysteretic operator was proposed to transform the multi-valued mapping into a one-to-one mapping which could enable neural networks to approximate the behavior of an inverse hysteresis.Simulation results showed that the proposed control strategy can adapt itself to changes of hysteretic characteristics of a GMA under different input reference signals,and an on-line inverse hysteresis model of a GMA can be obtained,thus the hysteretic impact can be eliminated and high precision control of a GMA can be achieved.