Resource-aware MPC for constrained nonlinear systems: A self-triggered control approach

Reducing usage of the computational resources by event driven approach to model predictive control

“…Easily can be seen that neither linear (q ) nor constant (r ) component is present in a MPC cost function (5).…”

“…Also an on-line approach for improving performance of the off-line control and resource schedule, called Optimal Pointer Placement , is presented in this work. Interesting approaches can also be found in the literature dealing with the networked control systems (NCS), two main approaches oriented towards achieving resource utilization reduction (to save computation and communication resources) can be distinguished, namely, Event-Triggered Control (ETC) and Self-Triggered Control (STC) [5,6]. The control law in ETC and STC consists of a feedback controller that computes the control signals, and a triggering mechanism that determines when the control signals have to be updated.…”

“…The difference between ETC and STC is that in the former the control update is triggered by specific condition which is continuously being checked and when it becomes true, the control signals are recalculated, while in the latter the next update time is determined at current update time. The paper [5] introduces a general framework for the selftriggered MPC strategy applying to discrete-time nonlinear systems subject to state and input constraints and possibly a non-quadratic cost function. The framework proposed in [5] also provides a priori closed-loop performance guarantees in terms of original cost function apart form asymptotic stability and constraint satisfaction.…”

“…The paper [5] introduces a general framework for the selftriggered MPC strategy applying to discrete-time nonlinear systems subject to state and input constraints and possibly a non-quadratic cost function. The framework proposed in [5] also provides a priori closed-loop performance guarantees in terms of original cost function apart form asymptotic stability and constraint satisfaction. If the messages in the communication network under consideration provide high space for useful data (payload), another mechanism to reduce utilization of communication resources (without resorting to sporadically chang-ing control signals) is at hand -ie packet-based predictive control.…”

Communication Scheduling in Robust Self-Triggered MPC for Linear Discrete-Time Systems

“…This implies that this particular schedule will be chosen for a smaller set of states in the state space. Define also V µ,M For any given x T k ∈ R n , T k ∈ N, the scheduling function Ω is defined by the optimization problem (12) which is adapted from Gommans and Heemels (2015). The intuition behind this definition of Ω is to select the schedule with the least number of required communications which guarantees (i) robust constraint satisfaction and (ii) a bound on the cost function which ensures the stabilization of a certain sequence of compact sets.…”