Asymmetric Hysteresis Modeling And Identification Of Hysteresis Nonlinearity System

Smart structures which are often composed of piezoceramic, gaint magnetostrictive, have some advantages of small size, fast response and high precision. They are widely used in the field of microelectronics manufacturing, ultra precision machining, aerospace etc. However, the hysteresis nonlinearity in these smart structures not only increases the difficulties of modeling, but also introduces oscillations and even instability. In recent years, many researchers have obtained rich results on the modeling of hysteresis and the identification of hysteresis nonlinearity system. One of the purposes is to describe the hysteresis nonlinearity more accurately for the enhancement of control performance.At present,symmetric hysteresismodeling is considered in the literatures, and the identification of hysteresis system is dependent on the measuring of intermediate variable.In this paper,asymmetric hysteresis modeling is considered based a Prandtl-Ishlinskii(PI)model, and a novel model identification method is developed for hysteresis systems based on extended state observer(ESO). The main contents and achievements are listed as follows.Firstly, an asymmetric PI hysteresis model is developed to describe the asymmetric hysteresis curves. The classicalPI model is divided into two portions,linear portion and operator summation. In order to enhance the symmetry of the rest curve of the hysteresis, a nonlinear input function is used in theasymmetricPI model as the center curve of hysteresis instead of the linear portion. The rest curve of the hysteresis is modeled by a new defined operator which provides some basic asymmetric hysteresis. The asymmetric PI model can be used to describe the symmetric as well as asymmetric hysteresis curves. A number of simulations are designed to verify the effectiveness of the asymmetric PI model for describe the asymmetric hysteresis.Secondly, a model parameter identification method based on ESO is developed for a class of Hammerstein-like nonlinear system, which is composed of a linear portion and a bounded hysteresis portion. A decomposition method to Bouc-Wenmodel is firstly presented with a centrosymmetric and periodical input, for the parameter separation of the linear potion and the hysteresis portion. Since all the information of the model can be estimated effectively as an ‘extended state’ by ESO,the parameters for linear portion and hysteresis portion are identified separately based on ESO. The simulation results are given to verify the effectiveness of parameter identification method.Finally, an experiment is carried on to identify the parameters of vehicle displacement system with hysteresis input, which is used to illustrate the effectiveness of the proposed method. Combined with a hysteresis input, the vehicle displacement system is used as a real hysteresis system. The model of hysteresis system identified by ESO is used to make forward inverse compensation in a classical PID feedback control loop. The experimental results show that the effectiveness of the proposed identification method.