Distributed model predictive control for constrained nonlinear systems with decoupled local dynamics

Area of Science:

• Control Engineering
• Systems Science
• Applied Mathematics

Background:

• Large-scale nonlinear systems present significant control challenges.
• Centralized model predictive control (MPC) can be computationally intensive for such systems.
• Distributed control strategies are needed to manage complexity and communication overhead.

Purpose of the Study:

• To develop a distributed model predictive control (MPC) strategy for nonlinear large-scale systems.
• To ensure closed-loop stability in a distributed MPC framework.
• To reduce communication requirements between subsystems.

Main Methods:

• Decomposition of the overall cost function into local optimization problems for each subsystem.
• Fixed neighbor identification based on coupled states in the cost function.
• Decomposition of centralized compatibility constraints into local controllers.
• Application of a quasi-infinite horizon MPC algorithm for each local controller.
• Analysis of communication protocols involving current states and optimized inputs.

Main Results:

• Successfully disassembled the global objective function into local optimization tasks.
• Ensured closed-loop stability by decomposing compatibility constraints.
• Demonstrated low communication overhead, transferring only essential state and input information.
• Proved exponential stability for the global closed-loop system.

Conclusions:

• The proposed distributed MPC approach is effective for nonlinear large-scale systems.
• The method guarantees system stability while minimizing communication.
• This framework offers a scalable and efficient solution for complex control problems.