Identification of piezoelectric hysteresis by a novel Duhem model based neural network

- The entire model is composed of the NN static submodel and the linear dynamic submodel.
- Duhem model is used to construct a neural network for hysteresis description and parameter identification.
- The result reveals the proposed approach can precisely model and identify the actuator response.

Duhem model has been widely used to describe the hysteresis property for many memory-type nonlinear systems. In this paper, a novel parameter identification method based on artificial neural network has been developed for the Duhem hysteresis model. With the Duhem differential equation, a neural network is constructed reasonably whose weights are specially designed mapping to the model parameters. Based on the universal function approximation capability of neural network, the model parameters can be identified with the proposed approach by network training. The parameter identification scheme is validated by simulation firstly and then applied to modeling the piezoelectric actuator. The results reveal that the proposed identification approach can identify the static hysteresis nonlinearity with high accuracy. Furthermore, combined with the dynamic component, the presented method can also be used to describe the dynamic hysteresis of the piezoelectric actuator.