Self-adjusted decomposition for multi-model predictive control of Hammerstein systems based on included angle

Area of Science:

• Chemical Engineering
• Control Systems Engineering
• Systems and Control Theory

Background:

• Hammerstein systems, characterized by a series of linear and static nonlinear elements, present significant challenges in control system design.
• Existing multi-model predictive control (MMPC) methods often require extensive tuning and rely heavily on empirical expertise for system decomposition.
• Accurate approximation of nonlinear system dynamics is crucial for effective model-based control strategies.

Purpose of the Study:

• To introduce a novel Self-Adjusted Multi-Model Decomposition (SAMMD) method for Hammerstein systems.
• To improve the efficiency and quality of decomposing Hammerstein systems into a set of linear models.
• To enhance the performance of Multi-Model Predictive Control (MMPC) through a more robust decomposition technique.

Main Methods:

• The proposed SAMMD method utilizes the included angle (IA) criterion for balanced decomposition based on the Measurement of Nonlinearity (MoN).
• An initial threshold value and step-size are provided to obtain an appropriate linear model set approximating the Hammerstein system.
• MMPC is designed using an offline weighting method on the obtained linear model set for set-point tracking and anti-disturbance control.

Main Results:

• The SAMMD method significantly reduces the need for time-consuming threshold tuning and reliance on user experience.
• The efficiency and quality of the system decomposition are substantially improved compared to traditional methods.
• Simulations on a Continuous Stirred-Tank Reactor (CSTR) and a Lab-tank system demonstrate the effectiveness and efficiency of the SAMMD-based MMPC.

Conclusions:

• The SAMMD method offers an effective and efficient approach for Hammerstein system decomposition.
• The proposed method enhances MMPC performance in set-point tracking and disturbance rejection for complex nonlinear systems.
• This work provides a valuable tool for improving the control of systems that can be modeled as Hammerstein structures.