Distributed Adaptive Neural Anti-Disturbance Cooperative Control of High-Order MIMO Nonlinear Multi-Agent Systems

In this paper, a distributed cooperative control problem for a class of high-order multi-input and multi-output (MIMO) nonlinear multiagent systems (MASs) in the presence of uncertain nonlinearities and external disturbances is addressed. A coupled design is developed to collaboratively approximate unknown nonlinearities and compounded disturbances by combining neural networks (NNs) with high-order disturbance observers (HODOs). To further simplify the controller structure, relationships among the Laplace matrix, adjacency matrix and consensus tracking errors are analyzed based on undirected communication graphs. Then, a distributed adaptive NN anti-disturbance control protocol is proposed for high-order MIMO nonlinear MASs based on the outputs of NNs and HODOs, where dynamic surface control (DSC) is introduced to eliminate the "computational explosion" problem of the conventional backstepping method. The semiglobally uniformly ultimate boundedness of closed-loop system signals is proven through Lyapunov theory. Finally, simulations of a quadrotor UAV formation are performed to demonstrate the effectiveness of the proposed control scheme.