Hysteresis Nonlinearity Modeling And Control Method For Piezoelectric Actuator Based On Hammerstein Model

Micro-nano actuator is the core of micro-nano positioning technology,so it is great significant to look for the excellent actuator material to improve the positioning accuracy.Piezoelectric ceramics is an intelligent actuator material whose energy form can be transformed between mechanical and electricity.And it has the superior points of small volume,low price,high displacement precision,etc.In recent years,it has become a hotspot and has been widely used in aerospace aviation,robotics,micro control engineering and other fields.However,there is a hysteresis phenomenon in the piezoelectric ceramic material,which leads to the slow response and the poor controllability of the piezoelectric ceramic actuator,which seriously hinders the application and development of the piezoelectric ceramic actuator(PEA).Therefore,it is crucial to how to eliminate the hysteresis phenomenon.The idea of eliminating hysteresis nonlinearity can be summarized as modeling and control.For the phenomenon of rate-dependent hysteresis nonlinearity in piezoelectric actuators,this paper first establishes the rate-dependent hysteresis nonlinear model,due to the traditional KP model belongs to the rate-independent hysteresis nonlinear model,so this paper proposes a kind of Hammerstein model based on the traditional KP model,which can describe the rate-dependent hysteresis nonlinearity in the PEA.This model consists of the classical KP model and the transfer function to cascade.The KP model is used to characterize the hysteresis nonlinearity.The transfer function is used to describe the dynamic rate dependency.Then,it adopts Harmony Search algorithm and the algorithm of the combination of Harmony Search and Bat to identify parameters.Finally,in order to reduce the influence of the hysteresis nonlinearity on the control of the displacement,this paper proposes three control schemes.One is the feed-forward control scheme based on the hysteresis inverse model.The experimental results show that the proposed feed-forward controller effectively achieves the compensation for the rate-dependent hysteresis nonlinearity of the PEA.Although the feed-forward control can improve the accuracy of tracking,but the ability of the system is too weak.Another is PID compound control scheme based on the inverse model,in which the parameters of PID are adjusting by BP neural network online.This scheme further improves the control precision and enhances the anti-jamming ability of the system.But the existence of the inverse model still weakens the effect of controller.The last control scheme is a non-inverse model of the control method.Based on the supernormal approximation ability to the nonlinear function of the neural network,two neural networks are used to approximate the two unknowns in the function relation between the output and input of the control system,so as to achieve the ideal accuracy of tracking.Experimental results show that the control effects of neural network are better than previous two control methods.