Research Of Control Method For Piezoelectric Ceramic Actuators Based On The Improved Krasnosel’skii-pokrovskii Model

Piezoelectric ceramic actuators have been widely used in the situation of microcosmicstructure or precision positioning by virtue of the property about generating a submicronresolution displacement. However, piezoelectric ceramic actuators which have somedifficulties, such as hysteresis, nonlinearity and creep reduce the output displacementaccuracy of the piezoelectric ceramic actuator. In order to reduce or eliminate the impact ofthese performances and enhance the displacement accuracy control of the piezoelectricceramic actuator, this paper is to design a new model of the piezoelectric ceramic actuatorand the control scheme based on this model.In this paper, first, the working principle, the performance and the main difficulties ofthe piezoelectric ceramic actuator during the research of the piezoelectric ceramic actuatorare introduced, we study and summarize the methods and achievements of the piezoelectricceramic actuator to overcome disadvantages of the piezoelectric ceramic actuator byscholars. The Krasnosel’skii-Pokrovskii (KP) model is definitely applied to establish thehysteresis nonlinearity model after comparing and analyzing pre-existing hysteresis models.Combining with the research in this paper, the KP model is modified by adding the inputvoltage change rate. After the hysteresis operator of KP model is confirmed it has able todescribe the dynamic characteristics of the piezoelectric ceramic actuator. Whereafter, theparticle swarm optimization (PSO) and linear time-delay neural networks (DTNN) are usedto identify the density parameters of KP model. Comparing the two results obtained by twoidentification methods, the new hysteresis model with higher accuracy and faster speed isobtained by the DTNN. Based on the hysteresis model of the piezoelectric ceramic actuator, the inversehysteresis nonlinearity model is proposed by the general regression neural network (GRNN).Then the inverse model is designed as the feedforward controller to compensate thepiezoelectric ceramic actuator. In order to overcome the shortcoming that the interference ishard eliminated by feedforward compensation control, the PID compound control schemebased on the genetic algorithm (GA) is proposed to adjust the parameters of the PIDcontroller. Because the maximum error rate of the piezoelectric ceramic actuator is reducefrom1.97%of the feedforward control scheme to1.18%of the compound control scheme,and it is proved in the simulation results that the compound control scheme is effective toimprove the accuracy obviously. What’s more, we take the shortage of unreal-timeperformance of the parameters of PID controller adjusted by the genetic algorithm intoconsideration, and we propose a compound control consisting of feedforward control andneural network PID control. The parameters of PID controller are adjusted in real time byneural network identifier so as to obtain a more adaptable system. The simulation resultsprove that this control scheme can control the displacement of the piezoelectric ceramicactuator more precisely, and the maximum error rate of the piezoelectric ceramic actuator isreduced to0.59%.At the end of this paper, a double closed-loop compound control scheme which theinverse model is used as a feedforward controller combining with neural network PID andadaptive inverse control is designed. It can not only reduce the errors of models and systemseffectively. The simulation experiments proved that, comparing with the neural networkPID compound control scheme, the double closed-loop compound control scheme canfurther reduce the hysteresis nonlinearity and improve the control accuracy of thepiezoelectric ceramic actuator because of its the maximum error rate is reduce to0.15%.