Stochastic sampled-data stabilization of neural-network-based control systems

This paper addresses the problem of stochastic sampled-data stabilization for neural-network-based control systems (NNBCSs) with an optimal guaranteed cost. In order to stabilize the closed-loop system, continuous-time nonlinear plant and three-layer fully connected feed-forward neural networks based on stochastic sampling are connected to the closed loop. By introducing new Lyapunov–Krasovskii functional with triple integral terms and by using second-order reciprocal convex technique, new stability and stabilization criteria for NNBCSs are derived in terms of linear matrix inequalities (LMIs). The desired stochastic sampled-data controllers can be calculated by solving these LMIs. Finally, physical example is given to verify the effectiveness and usefulness of the obtained results.