Event-Triggered Neural Network Control for LTI Systems

This letter addresses the problem of event-triggered control (ETC) discrete-time linear time-invariant (LTI) systems stabilized by neural network controllers. The event-triggering mechanism (ETM) is proposed to update only a portion layers required maintain stability and satisfactory performance feedback system, thus reducing computational cost associated with evaluation network. Sufficient convex conditions in form matrix inequalities (LMIs) are provided compute triggering parameters characterize an estimate domain attraction for system. formulation based on use Lyapunov theory set generalized sector constraints that deal nonlinear activation functions, represented this case saturation function. Optimization procedures also formulated effectively reduce amount computation An example borrowed from literature used illustrate effectiveness proposal.