Research On Hysteresis Characteristics Of Piezoelectric Ceramic Stage Based On Neural Network

In recent years,with the continuous development and progress in the field of ultraprecision machining,the role of micro-nano positioning technology has become more and more important.The choice of the core material of the actuator that produces micronano displacement has a very significant impact on the micro-nano positioning accuracy.As a typical smart material,piezoelectric ceramics have the advantages of fast response,high energy density,high resolution,and low quality,It can quickly convert electrical energy and mechanical energy into each other,it has been widely used in various micro-displacement drive control.The research of piezoelectric ceramic stage based on piezoelectric ceramic material has attracted more and more attention.However,due to the inherent hysteresis and nonlinear characteristics of piezoelectric ceramic materials,their positioning accuracy is greatly affected and the development of piezoelectric ceramic stages is hindered.Therefore,compensating the hysteresis and nonlinearity of piezoelectric ceramics has very important theoretical research significance and practical application value.This paper studies the hysteresis and nonlinear characteristics of piezoelectric ceramics.The main contents include:In order to solve the hysteresis nonlinear problem of the piezoelectric ceramic stage,aiming at the problem of the description accuracy of the traditional Backlash-Like model,an improved Backlash-Like model with improved dead zone operator structure and polynomial function introduced is proposed.Then,considering the influence of the increase of the input voltage leading to the increase of the modeling error,based on the traditional PI model,an improved segmented PI model based on unilateral play operator modeling was proposed.Compared with the Backlash-Like model,the segmented PI modeling can also fit the hysteresis loop of piezoelectric ceramics with higher accuracy under high-voltage input,and the modeling accuracy is greatly improved.Because the hysteresis nonlinearity of piezoelectric ceramics has dynamic characteristics,an RBF neural network model is established,and the input space is expanded by introducing hysteresis operators,and the multi-value mapping is transformed into a one-to-one mapping.Finally,through experimental comparison,it is shown that the segmented PI model and the Radial Basis Function(RBF)neural network model can characterize the hysteresis nonlinearity and dynamic characteristics of piezoelectric ceramics.The model identification is simple,the description accuracy is high,the frequency generalization ability is strong,and it has strong applicability.For the control of the piezoelectric ceramic stage,this paper proposes a coincidence control strategy based on the feedforward inverse model.The inverse model of the RBF neural network and the inverse model of the segmented PI model are established.The feedforward inverse model is used to compensate the hysteresis and nonlinearity of the piezoelectric ceramics,and the feedback link of the PID controller can greatly improve the stability of the control system.A related experiment platform was built,and a tracking control experiment based on sweep frequency signals was designed.The results show that the designed composite control system can carry out effective control,with high stability,small error,strong frequency generalization ability,and strong engineering practical value.