Event-Triggered Adaptive Horizon DMPC for Nonlinear Multi-Agent Systems: A Single/Dual-Mode Approach
Yu Jie Yang, Tianbo Zhang, Tao Wang, Dan Zhou, Dong Shen · 2025
In this paper, two event-triggered adaptive horizon distributed model predictive control (EADMPC) strategies are proposed for nonlinear multi-agent systems, subject to disturbances along with state and input constraints, and designed based on single-mode and dual-mode MPC schemes, respectively, aiming to address the computational and communication challenges inherent in DMPC implementations. In the single-mode scheme, an event-triggering mechanism is introduced based on state prediction deviations, with a single MPC controller applied throughout. The dual-mode scheme employs two distinct event-triggering mechanisms: one for the MPC controllers when the system states lie outside the terminal region, which aligns with the single-mode approach, and another for the local controllers when the states are within the terminal region, designed based on input-to-state stability conditions. Both strategies incorporate an adaptive prediction horizon update mechanism to further reduce computational complexity. Sufficient conditions to ensure recursive feasibility and closed-loop stability are rigorously derived, and simulations validate the effectiveness of the proposed algorithms.