H∞ Control of Markov Jump Reaction-Diffusion Neural Networks via a Dynamic Event-Triggered Mechansim with Actuator Faults

Source abstract

This paper investigates the problem of H∞ control of Markov jump reactiondiffusion neural networks (MJRDNNs) via a dynamic event-triggered mechanism with actuator faults. First, by introducing reaction-diffusion terms into neural networks with Markovian switching, a more applicable class is constructed. Considering the communication pressure of MJRDNNs, this paper adopts a dynamic event-triggered mechanism to make the model more practical. By introducing an internal dynamic variable, the event-triggered mechanism has a more flexible threshold. The potential actuator fault in each network node is also considered. Secondly, an integral feedback controller is presented to study the control problem and $\mathcal{H}_\infty$ performance of MJRDNNs. By constructing an integral-form Lyapunov-Krasovskii (L-K) functional, some sufficient conditions can be obtained. Moreover, by using matrix techniques, the sufficient conditions are transformed into solvable linear matrix inequalities (LMIs), and the controller gain can also be derived. Finally, a numerical example is presented to demonstrate the effectiveness of the results.