Stable neural network control scheme for companion-type nonlinear system

The companion-type nonlinear system, whose nonlinearities is unknown, was considered. The neural network was introduced as models of nonlinear functions. The control objective of the system is to make output of the system asymptotically tracking an expected value. Based on Lyapunov stability theory, the effective control law and adaptive law of parameters for the neural network model were proposed. On the basis of global invariant set theorem, the global tracking convergence of the close-loop system was proved. The simulation results have shown that the neural network control system is stable, and tracking accuracy will increase with the increase of number of neural elements in hidden layer, prolongation of tracking time and decrease of square sum.