Research On Modeling And Control Methods Of Hysteresis In Piezoelectric Ceramic Actuator Based On Neural Network

Piezoelectric ceramic actuators,as a kind of micro-precision components,have been widely used in various industry applications such as remote-controlled surgical robots,scanning probe atomic force microscopes,and gyros in aerospace flexible robots.However,the inherent hysteresis and nonlinearity existing in the piezoelectric ceramic actuators will not only affect the positioning accuracy of the system,but also cause the oscillation of the system,resulting in the performances of the entire positioning platform to degrade.At present,methods based on neural network modeling have great potential in nonlinear approximation and parameter identification.Therefore,it is of great significance to study how to compensate or suppress hysteresis through neural network methods for the application of piezoelectric ceramic actuators in micro-precision systems.According to the input and output data collected by the experimental platform,it is found that the hysteresis characteristics of the piezoelectric ceramic actuators change with the frequency of the input signal,which is called the rate-dependent hysteresis(i.e.,the dynamic hysteresis)in general.With the frequency of the input signal increasing,the width of the hysteresis loop increases and the height decreases.According to the characteristics of hysteresis,static model and rate-dependent model are established respectively,and based on the established model,the controller is designed to effectively compensate the influences of hysteresis nonlinearities in the system.The main research contents of this paper are listed as follows:(1)A new static hysteresis auxiliary operator with minor-loops congruence and wiping-out properties is proposed to construct the input space of the model,then a neural network hysteresis model based on the idea of expanded input space is established,and the feasibility of the model is proved.The proposed static hysteresis auxiliary operator also has a simpler structure.Afterwards,the experimental platform of the piezoelectric ceramic actuator is used to show the validation of the model.For comparison,the Preisach-type model is established too.The compared experimental results show that the proposed model is superior to the Preisach-type model in describing the hysteretic minor-loops congruence and wiping-out property.(2)When the frequency of the input signal changes larger,the static hysteresis model cannot accurately describe the hysteresis properties.Therefore,a new type of rate-dependent hysteresis auxiliary operator containing the frequency information of the input signal is proposed and the feasibility of expressing the rate-dependent hysteresis by the operator to a certain extent is analyzed.Then,based on the idea of expanded input space,a neural network model with rate-dependent hysteresis is built,and the mathematical analysis and proof are given.Finally,the validity of the raised model is verified by the experiments with piezoelectric ceramic actuator.(3)According to the established rate-dependent hysteresis model and its modeling ideas,the inverse model of the rate-dependent neural network model is constructed,and the hybrid controller of the piezoelectric ceramic actuator is designed.For comparison,the hybrid controller based on the MRPI inverse model and the PID controller are designed,respectively.Through comparison of experimental results,it is found that the performances of the presented controller are better than the other two controllers.