Predictor-Based Neural Dynamic Surface Control for Bipartite Tracking of a Class of Nonlinear Multiagent Systems

This article is concerned with bipartite tracking for a class of nonlinear multiagent systems under a signed directed graph, where the followers are with unknown virtual control gains. In the predictor-based neural dynamic surface control (NDSC) framework, a bipartite tracking control strategy is proposed by the introduction of predictors and the minimal number of learning parameters (MNLPs) technology along with the graph theory. Different from the traditional NDSC, the predictor-based NDSC utilizes prediction errors to update the neural network for improving system transient performance. The MNLPs technology is employed to avoid the problem of ``explosion of learning parameters''. It is proved that all closed-loop signals steered by the proposed control strategy are bounded, and the system achieves bipartite consensus. Simulation results verify the efficiency and effectiveness of the strategy.

Automated summary

This article proposes a control strategy for bipartite tracking in nonlinear multiagent systems by introducing predictors and minimal learning parameters technology, along with the use of graph theory. The strategy utilizes prediction errors to update neural networks and avoids the problem of learning parameter explosion, achieving bounded closed-loop control signals and bipartite consensus in the system. Simulation results confirm the efficiency and effectiveness of the strategy.