一种频率依赖性迟滞的神经网络内模控制

A Neural Network Internal Model Control of Rate-Dependent Hysteresis

针对压电陶瓷等智能材料存在的依赖输入频率的迟滞非线性问题,采用BP神经网络对迟滞非线性进行辨识,并通过内模控制方案来对其进行控制。在迟滞的建模上,构建了一种静态迟滞非线性环节串联一个对输入频率敏感的线性动态环节组成的Hammerstein模型。在此基础上,得出Hammerstein模型的逆模型,并通过构造的正、逆模型进行内模控制。实验结果说明,提出的建模方法与内模控制方案是有效的。

A BP neural network internal model control strategy is used to handle the influence of rate-depended hysteresis in piezoelectric material.In order to model the behavior of the rate-depended hysteresis,a Hammerstein model consisting of static hysteresis and a linear dynamic block is proposed.And then the inverse model of the hybird structure is obtained.The experiment results illustrate that the proposed model and its internal model control algorithm are effective.