基于混合神经网络的压电陶瓷微位移执行器动态迟滞建模

提出了两个动态神经网络串联的混合神经网络动态迟滞模型,用以逼近压电陶瓷的迟滞特性.混合模型由两个动态RBF神经网络构成,前者形成一个相位超前的动态模型,其特性与压电陶瓷的输出特性类似,但在相位和幅值上有所区别;后者实现相位滞后的变换和幅值的非线性变换,以达到对压电陶瓷实际输出的逼近.仿真和实验表明,所提出的描述动态迟滞特性的动态迟滞模型是有效的.与PI模型相比较,具有较高的模型精度.

Dynamic Hysteresis Modeling of Piezoceramic Actuator Based on Hybrid Neural Networks

The hybrid neural network dynamic hysteresis model, which consists of two dynamic neural networks in a cascade form,is proposed to approximate the hysteresis characteristics of piezoceramic actuator.The hybrid model consists of two RBF neural networks.The former is a dynamic model with the leading phase,the behavior of which is similar to the feaure of piezoceramic actuator,but differs from the feature of piezoceramic in the aspect of their phase and magnitudes.The latter is used to carry out the nonlinear transform of phase and magnitude for the approximation of output of piezoceramic actuator. Simulation and experiment on the piezoceramic actuator show that the proposed model to describe the behavior of piezoceramic actuator is effective.In comparison with PI model,the proposed model is of high precision.