Modeling And Control For Nonlinear Systems With Hysteresis

With the development of science and the in-depth understanding of humanstudies, conductive polymer and piezoelectric ceramics have found an increasinglywide utilization and higher operation accuracy in intelligent materials. This situationhas asked for higher requirement of the dynamic performance of the system. It hasexisted a nonlinear phenomenon caused by hysteresis that affected the dynamicperformance of the system in intelligent materials liked piezoelectric ceramics. Andthe phenomena has made it difficult to practical application of materials. The paperhas addressed this nonlinear phenomenon to study and established model forhysteresis nonlinear characteristics and compensation controlled.First of all, the hysteresis nonlinear characteristics of intelligent materials wereanalyzed and the model with hysteresis characteristics was established that was usedto simulate the hysteresis nonlinear phenomenon in intelligent material. The modelwas consisted of two parts in series. It included Preisach hysteresis model that was notaffected by the transfer function and the transfer function without considering theimpact of hysteresis.Secondly, identifying the system based on system model with hysteresischaracteristics. The identification method was divided into offline identificationmethod and online identification method. The offline used the least-square method，and the online used the recursive least squares method and neural networkidentification method. The least-square method was used to fit hysteresis systemmodel through the identification of the transfer function. The identificationparameters obtained by the least-square method was used as the online identificationof the initial value. The paper got the precise identification model by the recursiveleast squares method and neural network identification method. This way, thehysteresis system model was better fit. Thirdly, the compensation controller with hysteresis system model was builtbased on hysteresis identification model. Four kinds of controllers were builtaccording to the hysteresis system model. There were the inverse model controllers,PID inverse model controller, the neural network inverse model controller and theneural network inverse model controller based on hysteresis model identification. Thefirst two controllers was built by the identification of the transfer function that wasobtained by discrete identification method. The latter two controllers was built byidentification model that was obtained by online identification.Finally, the simulation results have showed that the identified model and thecontrollers established on the nonlinear hysteresis characteristics were feasible andeffective. These models and controllers could compensation control the hysteresisnonlinear phenomenon ideally.