Two-stage observer based offset-free MPC.

Area of Science:

• Control Engineering
• Systems Science
• Optimization Theory

Background:

• Model predictive control (MPC) is widely used for dynamic systems.
• Disturbances and model mismatch pose significant challenges to control performance.
• Observers are typically employed to estimate states and disturbances for rejection.

Purpose of the Study:

• To extend model predictive control (MPC) for dynamic systems subject to disturbances.
• To propose a novel two-stage design approach utilizing a full state and disturbance observer.
• To leverage the separation principle for independent observer and MPC design.

Main Methods:

• A two-stage approach is proposed, decoupling observer design from the MPC problem.
• Observer design is achieved via a norm minimization problem to mitigate disturbance effects.
• Model predictive control (MPC) law is derived using multi-parametric quadratic programming.

Main Results:

• The observer design is successfully decoupled from the MPC problem, adhering to the separation principle.
• Independent observer design effectively reduces the impact of disturbances and model mismatch.
• The proposed method enables robust control for dynamic systems with disturbances.

Conclusions:

• The presented two-stage approach provides an effective framework for designing model predictive control for disturbed systems.
• Decoupling observer design simplifies the overall control system design process.
• The method enhances disturbance rejection capabilities and overall system robustness.