Configuration and reduced-order modeling of a flow system based on experimental data

Area of Science:

• Engineering
• Control Systems
• System Dynamics

Background:

• Regulated flow systems display variable dynamics and non-linear behavior, especially at low inputs.
• Modeling these systems is challenging due to differing output patterns during flow increases versus decreases.
• Developing separate models for each input-output set is computationally intensive for broad control applications.

Purpose of the Study:

• To develop a computationally efficient modeling technique for regulated flow systems.
• To create a unified dynamic model applicable across the entire input range.
• To enable the design of a single controller for comprehensive system management.

Main Methods:

• Utilized Principal Component Analysis (PCA) for parameter reduction.
• Developed a Linear Parameter Varying (LPV) dynamic model based on laboratory flow system data.
• Implemented automated data acquisition and model estimation.
• Reduced dynamic parameters using PCA to create a reduced LPV model.

Main Results:

• The developed LPV model accurately represents laboratory flow system dynamics.
• The model's output response closely matches experimentally measured flow rates.
• PCA effectively reduced the complexity of dynamic parameters.

Conclusions:

• The PCA-based LPV modeling technique offers an efficient solution for complex flow systems.
• This approach facilitates the design of a single controller effective over the entire operating range.
• The method shows promise for synthesizing advanced LPV controllers for flow control applications.