Input history dependent Preisach model of piezoceramic actuators

The hysteresis of piezoceramic actuators is the main challenge of its positioning accuracy. The outputs of the classical Preisach model describing the hysteresis nonlinearity are closely related with the input history. By introducing the input history values into the Preisach weight function, a new Preisach model was established in this paper. Then a neural network was built to identify the new model based on simplified hysteresis operators. The experimental results showed that when the model was excited by the exponential decayed sinusoidal voltages, the maximum of the absolute value of the displacement error predicted by the new model was reduced by 76% compared with that of the classical Preisach model, and the mean square error is reduced by 78%.