Distributed adaptive neural control of nonlinear multi-agent systems with unknown control directions

In this paper, the problem of distributed adaptive neural control is addressed for a class of uncertain non-affine nonlinear multi-agent systems with unknown control directions under switching directed topologies. Via mean-value theorem, non-affine follower agents’ dynamics are transformed to the structures so that control design becomes feasible. Then, radial basis function neural networks are used to approximate the unknown nonlinear functions. Due to the utilization of a Nussbaum gain function technique, the singularity problem and requirement to prior knowledge about signs of derivative of control gains are removed. On the base of dynamic surface control design and minimal learning parameter approach, a simplified approach to design distributed controller for uncertain nonlinear multi-agent systems is developed. As a result, the problems of explosion of complexity and dimensionality curse are counteracted, simultaneously. By the theoretical analysis, it is proved that the closed-loop network system is cooperatively semi-globally uniformly ultimately bounded. Meanwhile, convergence of distributed tracking errors to adjustable neighborhood of the origin is also proved. Finally, simulation examples and a comparative example are shown to verify and clarify efficiency of the proposed control approach.