| Piezoelectric ceramics are the ideal driving components for the current micro-positioning system,but their hysteresis,creep and other mechanical characteristics have seriously affected the accuracy and stability of their output.This article takes piezoelectric ceramics as the research object,deeply studies the hysteresis modeling method and compound control scheme based on Duhem model,and conducts experiments.In this paper,the causes of hysteresis and creep of piezoelectric ceramics are studied from a microscopic perspective,and the general laws of hysteresis and creep of piezoelectric ceramics are obtained through experiments.The unknown part of the Duhem model is replaced by a polynomial.According to the input and output data of different frequencies and amplitudes,the inverse model is obtained by recursive least squares identification.Since the input voltage value at the current moment cannot be predicted,the above method has a large error when calculating the reverse output.Therefore,this paper combines the Duhem inverse operator with the GRNN neural network to avoid inaccurate calculations in the inverse process,and directly establishes the neural network hysteresis inverse model.Compared with the above modeling techniques,the prediction output accuracy of the inverse model is improved by 41.7%.An adaptive sliding mode controller is designed,and Lyapunov stability analysis and simulation experiment verification are given.Through the hysteresis suppression,creep suppression and displacement tracking experiments,the correction capability of the controller designed in this paper is verified.The experimental results show that under compound control,the amount of hysteresis is reduced by 96%,and when the step amplitude is 1V,the creep is reduced by 49.1%.Under different frequency inputs,the maximum error of displacement tracking is reduced by 25.7% and the average absolute error is reduced by 17.5% compared with direct control. |
PDF Full Text Request