Abstract: The dynamic neural network hysteresis model of the piezoceramic actuator is presented by using the similar diagonal current neural network. This model is of structure of the Preisach model of which hysteresis cell is replaced by first-order differential equation. Simulation and experimental results performed on the two kinds of decayed rectangular input signals show that this new approach of identification and prediction for the piezoceramic actuator is effective and can achieve high accuracy.

Key words: First-order differential equation, Hysteresis behavior of the piezoceramic actuator, Preisach model, Similar diagonal recurrent neural network