神经网络广义逆系统控制

Generalized ANN Inverse Control Method

中文关键词  神经网络  逆系统  非线性系统控制  线性化  解耦

英文关键词  artificial neural networks  inverse system  nonlinear system control  linearization  decoupling

中文摘要

      提出适合于高阶非线性系统线性化解耦的广义逆系统. 它与被控系统复合后, 不但能实现原系统的线性化和解耦, 而且通过合理地设计逆系统, 可使伪线性复合系统的极点在复平面上任意配置. 进一步提出由静态神经网络和若干积分惯性等线性环节组成的神经网络广义逆系统, 为模型未知且内部状态不易测量的高阶非线性系统的线性化解耦控制提供一条有效途径, 进一步拓展了神经网络逆系统控制方法的适用范围.

英文摘要

      The paper presents extended inverse method for the linearization and decoupling control of high_order nonlinear system. Being rightly designed, generalized inverse can transform the controlled nonlinear system into a number of pseudo_linear SISO (single input and single output) subsystems with the poles of the pseudo_linear systems approaching to the expected positions. Furthermore, the paper proposes the construction of generalized ANN (artificial neural network) inverse consisting of a static ANN and a number of linear components such as integrator factors and inertial components. This generalized ANN inverse method provides a practical approach for the linearization and decoupling control of poorly_modeled high_order plants whose states are difficult to measure and so extends the application region of ANN inverse.