Neural network control of nonlinear time-delay system with unknown dead-zone and its application to a robotic servo system

An adaptive controller is proposed for a class of nonlinear systems with unknown time-varying delays and a dead-zone input. Taking the dead-zone as a part of the system dynamics, the construction of the dead-zone inverse model is not needed and thus the characteristic parameters of the dead-zone are not necessarily known. Unknown time delays are handled by introducing improved Lyapunov-Krasovskii functions, where the requirements on the delayed functions/control coefficients are further relaxed without the singularity problem. A novel high-order neural network with only a scalar weight parameter is developed to approximate unknown nonlinearities. The closed-loop system is proved to be semi-globally uniformly ultimately bounded (SGUUB). Experiments on a robotic servo system are provided to verify the reliability of the presented method.