NN-Based Dynamic Adaptive Triggered Security Compensation Feedback Control for Implicit Power Systems Subjected to FDI Attacks

This paper investigates neural network-based dynamic adaptive event-triggered security compensation feedback control for implicit power systems (IPSs) subjected to false data injection (FDI) attacks. By introducing a time-varying positive exponential term, a new dynamic adaptive event-triggered mechanism (DAETM), which can avert Zeno phenomenon and address the issue that the controller fails to cancel the internal impulses of singular systems when using period-sampling based event-triggered mechanisms, is proposed to achieve the goal of enhancing communication efficiency while relieving channel load. Via adopting radial basis function (RBF) neural network (NN) approximation method to estimate the unknown FDI attack signals, the event-based security compensation feedback (SCF) controller with adaptive corrective term is designed, which can compensate the FDI attack signals and minimize the effect of FDI attacks. Considering the designed DAETM as well as the adopted RBF NN approximation method, a novel Lyapunov-Krasovskii (L-K) functional is constructed, which captures the dynamic adaptive event-triggered information and facilitates to obtain the ideal adaptive law of NN weight matrix, thereby contributing to the realization of regularity, impulse-freeness and ultimate boundedness of IPSs and promoting the collaborative design of DAETM and SCF controller. The final simulation results for IPSs validate the effectiveness of the studied methodology.