Generalized ANN Inverse Control Method

何丹,戴先中,王勤

DOI: https://doi.org/10.3969/j.issn.1000-8152.2002.01.006

2002-01-01

Abstract: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.