Dynamic event-triggered H∞ state estimation of switched NNs subject to ADT constraint and time-varying measurement delay

This paper addresses the non-fragile H∞ state estimation issue of switched neural networks subject to time-varying measurement delay under average dwell-time (ADT) constraint. For the sake of lowering communication overhead and boosting resource efficiency, an innovative dynamic event-triggered approach is proposed. This mechanism adaptively adjusts the triggering threshold through internal dynamic variables, effectively avoiding the Zeno phenomenon and reducing unnecessary signal transmission. Meanwhile, considering the case where the estimator gain has norm-bounded uncertainty, a non-fragile estimator with robust performance is introduced. By constructing a mode-dependent Lyapunov–Krasovskii functional and combining ADT framework with improved matrix inequality techniques, several sufficient criteria have been established to guarantee error system achieves expected exponential stability and meets the desired H∞ disturbance performance. On this basis, the expected estimator gain matrices can be effectively designed. At the end, one numerical example is proposed to verify the feasibility of designed estimation strategy. Meanwhile, simulation part also quantifies the specific impact of involved parameters on the system performance.