Adaptive Neural-Network-Based Control for a Class of Nonlinear Systems With Unknown Output Disturbance and Time Delays

This paper pays close attention to the adaptive neural network tracking control. Aiming at a class of uncertain nonlinear systems with completely unknown output disturbance and unknown time delay, a corresponding robust control method is proposed based on the backstepping design technology. Neural network approximation is introduced as a very effective estimation technique for modeling uncertain partitions in the design process of virtual controller. The suitable Lyapunov-Krasovskii function is constructed, and by using the organic combination of Young's inequality, unknown time delays are compensated. Nussbaum function is used to handle unknown virtual control directions. A practical robust control method is proposed to deal with the controller singularity problems. A priori knowledge is not required for this method. In this method, all signals achieve semi-global uniform ultimate boundedness, and it is demonstrated that the tracking error eventually converges the region around the origin. The simulation results verify this method's feasibility and effectiveness.