This paper investigates the sampled-data \(H_{\infty }\) control for a class of nonlinear fractional reaction-diffusion neural networks (FRDNNs) with time-varying delays and exogenous disturbances by output feedback. Herein, diffusion coefficients are the spatially varying matrix. For such system, sensors are assumed to provide point measurements (PMs) or averaged measurements (AMs), which are supposed to be perturbed. The first task is to design an observer to estimate the system state under PMs or AMs. Then, based on the Lyapunov method and Wirtinger’s inequality, sufficient conditions for asymptotic stability together with \(H_{\infty }\) performance of observer error dynamics are derived, where the Razumikhin theorem is applied to address time-varying delays. Secondly, with the proposed output feedback controller characterized as sampled-data in space and continuous in time, the \(H_{\infty }\) control problem of the closed-loop system is solved. Finally, numerical simulations are conducted to demonstrate the validity and feasibility of our results.
