Command-filter based predefined-time neural adaptive decentralized control for interconnected systems with dynamic uncertainties

This article studies the predefined-time controller construction problem for non-strict feedback nonlinear large-scale interconnected systems including unmodeled dynamics and dynamic disturbances. The uncertain nonlinearities and interconnections in the studied systems are universally approximated by neural networks. The observable dynamic signals created by the constructed auxiliary systems are utilized to alleviate the impacts of unmodeled state dynamics. Then, a neural adaptive predefined-time decentralized controller is developed via command filtered backstepping technology. The convergence time bound by the proposed control strategy is more flexible than conventional fixed-time control method. Moreover, through adding the compensation signal terms to the entire Lyapunov energy function, all variables in the controlled system are proved to be predefined-time bounded. The simulation shows the validity and superiority of the presented controller.