Dynamic model of hysteresis for piezoelectric actuators based on hysteretic operator and neural networks

In order to compensate the effect of hysteresis in piezoelectric actuators, a dynamic hysteresis model based on neural networks is proposed. In this method, a dynamic hysteretic operator is introduced to extract the memory property and rate-dependent property of hysteresis. The non-smooth property is also described in the operator. Moreover, the multi-valued mapping of hysteresis is decomposed into a one-to-one mapping which enables neural networks to approximate the behavior of hysteresis. Thus, the dynamic hysteresis model based on neural networks is derived. The proposed neural model is simple in structure and available for rate-dependent hysteresis. The weights of neural networks can be adjusted to adapt for different conditions. Finally, the proposed approach is applied to model the hysteresis in piezoelectric actuators.