基于退化激励信号的含有迟滞的三明治系统辨识

含有迟滞的三明治系统不仅具有非光滑、多值映射等特性, 而且迟滞环节的输入输出信号还是不能直接测量的, 常规方法难以进行有效的辨识. 本文提出了一种基于退化激励信号的两步辨识方案: 第一步, 设计一个特殊的退化激励信号将迟滞环节退化为一条静态曲线, 从而可以将两端的线性动态环节辨识出来, 解决中间信号不可测的问题; 第二步, 利用已辨识的线性模型重构迟滞环节的输入输出信号, 再采用“扩展输入空间法”建立迟滞环节的神经网络模型. 最后, 在压电超精密运动系统的实验结果表明所提出的建模方法取得了令人满意的结果.

Identification of sandwich systems with hysteresis based on degeneration inputs

Because of the nonsmooth nonlinearity in multivalued mapping, it is difficult to apply the conventional identification method to identify the model of hysteresis. If the hysteresis exists between two linear subsystems, it is more challenging to identify this sandwich system because both the input and output of the hysteresis cannot be measured directly. To deal with this problem, we propose a two-stage method based on degeneration inputs. A special exciting signal called the degeneration input is designed to degenerate the hysteresis into a static smooth function with one-to-one mapping, so that the linear subsystems can be estimated. Then, based on the obtained linear submodels, both the immeasurable input and output of hysteresis are reconstructed and a neural-network-based model of the hysteresis can be obtained by using the expanded input space method. Experimental results on a piezoelectric positioning system are presented to illustrate the performance of the proposed identification scheme.