Decentralized adaptive control of a class of mechanical large-scale nonlinear systems using neural networks

For a class of large-scale mechanical nonlinear systems, such as inverted pendulum, intervehicle spacing regulation in a platoon of automated highway systems and single-link robots with flexible joints, in this paper, we design a neural network (NN-based) decentralized adaptive control scheme to implement control them effectively. As a direct adaptive controller, neural network is used to approximate the unknown function of subsystem. Two designed robust control terms are applied to shield interconnection effect and disturbances, respectively. The stability is proved rigidly. The effectiveness is validated through simulation.