人工神经网络在解耦控制中的应用

针对强耦合带延时输入输出（MIMO）非线性离散系统难以实现解耦的问题，提出了用人工神经网络对未知非线性函数具有学习和逼近的能力，并与逆系统相结合，以构造逆系统函数来实现对不满足条件非线性离散系统线性化解耦。该人工神经网络系统由单个静态的神经网络和若干个时延因子构成，其神经网络仅用来逼近系统的静态函数，时延因子用来表征逆系统的动力学特征。这种逆系统可使系统结构简单，具有连好的动态性能，并以仿真实例给予了验证。

The Application of ANN in Decoupling Control

Aiming at the difficult decoupling problem in non-linear discrete system with time-delaystrong coupling and MIMO, by using ANN with the learning and imminent capability toward an unknown non-linear function and the combination with the inverse-system, the constructing a inverse-system functionto realize linearization decoupling of the non-linear discrete system with the dissatisfactory conditions ispresented. This ANN system is made up with a single static ANN and some time-delay factors. ANN approaches the system static function, and time-delay factors show the kinetic characters of inverse-system.The inverse-system makes structure simple and kinetic character better. Moreover, the ANN system isvalidated with the sample for emulate in this paper.