基于神经网络的迟滞逆模型

A neural-network-based inverse hysteresis model

一个新的基于神经网络的迟滞逆模型被提出.采用连续坐标变换的方法,建立基本迟滞逆算子(EIHO),EIHO为神经网络提供了基本的迟滞逆信息,并与迟滞逆的输入一起作为神经网络的输入,使迟滞逆由多值映射关系转化为一对一映射关系,从而达到用神经网络逼近迟滞逆的目的.一组实测数据被用来检验模型有效性,实验结果表明,这种建模方法是有效的.

A new neural-network-based inverse hysteresis model is proposed in this paper. The continuous transformation technique is used to construct an elementary inverse hysteresis operator （EIHO）, which extracts the elementary information of inverse hysteresis. The output of the EIHO is then used as one of the input signals of the neural network （NN） so that the multi-valued mapping of inverse hysteresis is transformed into a one-to-one mapping. In this way, neural networks can be used to model inverse hysteresis. A set of real data is also used to validate the effectiveness of the proposed approach. Finally, simulation results indicate that the proposed approach is successful.