An efficient algorithm for collaborative learning model predictive control of nonlinear systems.

Area of Science:

• Control Engineering
• Artificial Intelligence
• Systems Science

Background:

• Model Predictive Control (MPC) is crucial for complex systems but faces computational challenges.
• Collaborative control aims to improve system performance by coordinating multiple subsystems.
• Learning-based MPC integrates data-driven approaches to adapt control strategies.

Purpose of the Study:

• To develop an efficient computational algorithm for collaborative learning model predictive control (CLMPC) in nonlinear systems.
• To explore how subsystems can collaboratively achieve control objectives under constraints.
• To reduce the computational burden of MPC through a modified barycentric function.

Main Methods:

• Implementing a collaborative approach within the learning model predictive control framework.
• Modifying the barycentric function to decrease computational complexity.
• Proving theoretical properties such as recursive feasibility, stability, convergence, and optimality.

Main Results:

• The proposed CLMPC strategy effectively reduces computational load.
• The algorithm demonstrates recursive feasibility, stability, and convergence.
• Optimality of the control trajectory is maintained under collaborative operation.

Conclusions:

• The developed CLMPC algorithm offers an efficient and robust solution for nonlinear systems.
• Subsystem collaboration is a viable strategy to enhance MPC performance and reduce computational demands.
• The theoretical properties ensure reliable system operation and trajectory tracking.