Event-Triggered Active MPC for Nonlinear Multiagent Systems With Packet Losses.

Area of Science:

• Control Systems Engineering
• Networked Systems
• Nonlinear Dynamics

Background:

• Multiagent systems (MAS) are crucial in distributed control but face challenges like packet losses.
• Traditional control methods can be computationally intensive and require frequent data exchange.
• Event-triggered mechanisms offer a way to reduce communication and computational load.

Purpose of the Study:

• To develop an event-triggered active model predictive control strategy for nonlinear MAS with packet losses.
• To reduce sensing costs and ensure synchronous updating of control parameters.
• To address the impact of packet losses on system performance.

Main Methods:

• Designing event-triggered mechanisms to detect control update conditions at specific sampling instants.
• Actively selecting prediction horizons for individual agents via event-triggered conditions.
• Establishing a common maximal predictive horizon for synchronous updating across the MAS.
• Utilizing Bernoulli distributions to model packet losses within the model predictive control framework.

Main Results:

• The proposed event-triggered approach effectively manages nonlinear MAS under packet loss conditions.
• Synchronous updating of a common predictive horizon was achieved, enhancing system coordination.
• Reduced sensing costs were demonstrated through the event-triggered mechanism.
• Numerical simulations verified the effectiveness and robustness of the developed control algorithm.

Conclusions:

• The event-triggered active model predictive control is a viable and effective strategy for nonlinear MAS with packet losses.
• The method offers advantages in terms of reduced communication overhead and computational burden.
• The proposed approach ensures reliable and synchronized control in the presence of communication uncertainties.