Dynamic memory event-triggered interval type-2 fuzzy RBFNN control for multi-agent systems under false data injection attacks

This paper investigates an adaptive neural network event-triggered control problem for cooperative-competitive non-affine multi-agent systems (MASs) with input saturation under false data injection (FDI) attacks. Different from existing results, the two-channel smooth dynamic memory event-triggered mechanism is proposed, which not only stores historical information of dynamic variables to increase the triggering interval, but also uses smooth segmentation functions to circumvent the limitations caused by discontinuous trigger signals. Furthermore, the interval type-2 fuzzy radial basis function neural network with lower and upper membership functions is applied to MASs under FDI attacks, which relaxes the deterministic constraints on membership functions thus effectively handling nonlinear features. On this basis, a new coordinate transformation and auxiliary function are designed to mitigate the effects of FDI attacks and input saturation on system robustness, respectively. Finally, the Lyapunov stability theory is used to demonstrate the bipartite consensus performance of MASs, and all signals in the closed-loop system are bounded. The comparative simulation experiment of MASs in multiple RLC circuit systems verifies the rationality of the proposed method.