Adaptive Neural Network Fixed-Time Leader-Follower Consensus for Multiagent Systems With Constraints and Disturbances

This article is concerned with fixed-time leader-follower consensus problem for multiagent systems (MASs) with output constraints, unknown control direction, unknown system dynamics, unknown external disturbance, and dead-zone control input. First, a fixed-time distributed observer is presented for each follower to estimate the leader's states. Next, using a modified nonlinear mapping, an output-constrained system is transformed into an unconstrained system. Then, by adopting adding a power integrator technique, radial basis function neural network (RBFNN) approximation, and adaptive method, the ideal fixed-time stable virtual control protocol is derived and the issues of unknown control direction, unknown system dynamics, and unknown external disturbance are addressed. Finally, the actual control protocol is developed using the bound of dead-zone parameters. It is shown that the proposed control scheme achieves fixed-time leader-follower consensus of the studied MAS. The presented control protocol is applied to the leader-follower consensus of inverted pendulums and simulation results verify its effectiveness.

Automated summary

This article discusses the fixed-time leader-follower consensus problem for multiagent systems and addresses challenges such as output constraints, unknown dynamics, and unknown disturbances. By introducing distributed observers, nonlinear mappings, RBFNN approximation, and adaptive techniques, the ideal fixed-time stable virtual control protocol is derived to achieve consensus within a specified time frame. The proposed control scheme is successfully applied to inverted pendulums, demonstrating its effectiveness through simulation results.