Adaptive NN Consensus Control for Second-Order Nonlinear Multi-Agent Systems Against Sparse Sensor Attacks

This paper investigates the secure consensus tracking control problem of second-order nonlinear multi-agent systems against sparse sensor attacks. A selector of secure data is designed to filter out unattacked sensor output data from a set of output measurements affected by sensor attacks. Subsequently, a neural network (NN) secure state observer is then suggested to estimate the unavailable states of the system using the unattacked output data. Then, an adaptive NN consensus controller is proposed via dynamic surface control technique. The designed control method for the closed-loop systems ensures that all signals are ultimately bounded, and consensus tracking control is achieved with bounded tracking errors in the presence of sparse sensor attacks. Finally, the effectiveness of the proposed control scheme is validated through a simulation of unmanned aerial vehicle attitude control systems.