New kind of generalized Preisach hysteresis model and its identification based on neural network

The multi-valued-mapping hysteresis nonlinearity of the piezoceramic actuator seriously degrades the scale accuracy when used in the nanometer positioning system. Based on the classical Preisach model, this paper presents a kind of generalized Preisach model by introducing an average hysteresis function. Then the neural networks are used to identify the key parameters of the model. The results of simulation showed that the generalized Preisach model seemed to be more accurate than the classical Preisach model.