Research On Control Method Of Magnetic Shape Memory Alloy Actuator Based On U Model

Magnetic Shape Memory Alloy(MSMA)is a new type of intelligent material with Magnetic Shape Memory effect,which has the advantages of high displacement resolution,high energy density and long travel,and has a broad application prospect in the fields of highprecision positioning technology,precision manufacturing and biomedical engineering.MSMA actuator under the action of magnetic field using the magnetostrictive properties of MSMA material can produce micro and nano level of output displacement,can be used in the field of high precision positioning,with incomparable advantages of traditional actuators and other intelligent material actuators.However,due to the non-differentiable and non-one-toone mapping of the input and output of the MSMA actuator,the complex dynamic hysteresis nonlinear characteristics make the modeling and control of the system more difficult,and hinder the application of the MSMA actuator in the field of high precision positioning.In this paper,hysteresis nonlinearity is eliminated by establishing accurate models and designing reasonable control methods,so as to improve the control accuracy of the system.The main research contents are as follows:Firstly,on the basis of analyzing the characteristics of hysteresis ring and the hysteresis characteristics of MSMA actuator such as rate dependence and force dependence,a new model U model with simple structure is introduced to describe the dynamic hysteresis nonlinear of MSMA actuator.The U model is composed of two parts in series,one is a U static hysteresis model identified by the diminishing memory least square method to describe the static characteristics of hysteresis nonlinearity of MSMA actuator.The other part is a Wavelet Neural Network(WNN)which uses Nesterov Accelerated Gradient(NAG)to describe dynamic hysteresis characteristics.Experimental results show that the proposed model can accurately describe the dynamic hysteresis nonlinearity of MSMA actuator and has good dynamic performance.Then,a neural network control method based on U model is proposed for the influence of the hysteresis characteristic of MSMA actuator on the trajectory tracking control accuracy.Compared with the traditional inverse compensation open-loop feedforward control,the method eliminates the inverse calculation process and gets rid of the dependence of the control method on the inverse model.In order to further improve the control accuracy,a Whale Optimization Algorithm(WOA)is introduced to optimize the initial weights and thresholds of WNN on the basis of the neural network control method based on U model.Experimental results show that the optimized WOA-WNN control method based on U model has higher control accuracy and faster convergence speed,and can effectively avoid the inherent defect of WNN easily falling into local optimum.Finally,a neural network Iterative Learning Control(ILC)based on Dynamic Expansion Compression Coefficient(DECC)is proposed.ILC method improves control performance of MSMA actuator under high frequency input and high load.This control method combines ILC and neural network to overcome the shortcomings of traditional ILC,such as fixed parameters and strict repetition of initial state of iteration.DECC improves the accuracy of the iterative algorithm and expands the convergence range of the algorithm.This paper proves the convergence of ILC method based on constraint conditions.The experimental results verify that the neural network ILC introduced with DECC can effectively solve the control accuracy problem of WOA-WNN control algorithm under high frequency and high load conditions,and improve the overall control performance.