Data-driven abstractions for periodic event-triggered control systems based on informativity analysis

This paper proposes a data-driven abstract modelling approach for Periodic Event-Triggered Control (PETC) systems with unknown dynamics, from a finite number of system data. The construction approach consist of two steps. Firstly, finite input–state data are utilised for informativity analysis, and a discrete-time linear system is generated, whose set of parameter matrix pairs is estimated by this analysis. In this way, the inter-event dynamics are captured. Then, an abstract model is built from this discrete-time system, with the following procedure: firstly partition the state space, secondly compute the output map, thirdly find the transition relation. Thus, the dynamics of the original system can be captured by the constructed abstract model. Finally, a numerical example on a linear time-invariant system demonstrates that the proposed method can accurately approximate the dynamic behaviour of unknown systems from finite system execution data, validating its effectiveness and practical applicability.