Skip to Main Content
In this paper, for a class of nonlinear multivariable discrete-time systems, an open-loop approximately dynamical decoupling control law is first presented. Then, by introducing a lambda o T difference operator, an intelligent decoupling control method using multiple models and neural networks (NNs) is developed. The intelligent decoupling control method includes a set of fixed decoupling controllers, a reinitialized NN adaptive decoupling controller, a free-running NN adaptive decoupling controller, and a switching mechanism. Theory analysis shows that the free-running NN adaptive decoupling controller can guarantee the bounded-input-bounded-output stability of the closed-loop system, while the multiple fixed decoupling controllers and the reinitialized NN adaptive decoupling controller are used to improve the system performance. To illustrate the method, the proposed design is applied to a 2.4 times 2.4-m injector-driven transonic wind tunnel system. Simulation and industrial experiment results show the effectiveness and practicality of the proposed method.