Data-driven Intermittent Control with Neural Feedback Loop Design for Linear Systems

This paper proposes a data-driven intermittent control method based on a feed-forward Neural Networks (NNs) controller design and stability guarantees for plants with unknown dynamics. Then state-dependent regions are given and the global stability of the derived state-dependent intermittent control system is analyzed in detail, and the parameters in the pre-given state-dependent regions are adjustable, so the design is more flexible. It is worth making the point that we do not have to transmit real-time status measurements of regions where non-control is applied, which would reduce the update rate and computational complexity of NNs. The data storage and transportation costs are also saved. Finally, the control performance of the method is verified by a numerical example.