Neural-Network Decoupled Sliding-Mode Control for Inverted Pendulum System with Unknown Input Saturation

Abstract

In this paper, a neural-network decoupled sliding-mode control (NNDSMC) scheme is proposed for inverted pendulum system with unknown input saturation. The input saturation is approximated by a smooth affine function according to the mean-value theorem. By decoupling the whole inverted pendulum system into two second-order subsystems, two sliding manifolds are designed for each subsystem, in which the first sliding manifold includes an intermediate variable related to the second one. Finally, a nonsingular terminal sliding-mode control is employed for both subsystems by using a simple sigmoid neural network to approximate the unknown system nonlinearity. Simulations show the effectiveness of the presented method.