Generalized minimum variance control under long-range prediction horizon setups

Area of Science:

• Control Engineering
• Process Control
• Systems Engineering

Background:

• Generalized Minimum Variance (GMV) control is a widely used technique for process control.
• Stochastic disturbances and oscillations can degrade the performance of control systems.
• Existing methods may have limitations in handling long-range disturbances or require complex design procedures.

Purpose of the Study:

• To design and evaluate a minimal order Generalized Minimum Variance controller with a long-range prediction horizon.
• To investigate the impact of an increased prediction horizon on controller and plant output variances.
• To assess the controller's ability to mitigate stochastic disturbances and attenuate oscillations.

Main Methods:

• A novel design procedure for high-order prediction minimum variance filters, independent of the Diophantine Equation, was developed.
• The controller was evaluated using simulations and practical experiments on a first-order water flow rate system and a second-order under-damped electronic circuit.
• An incremental control scheme and identified stochastic models were employed for assessment.
• Two optimal tuning procedures for the algorithm were proposed.

Main Results:

• The minimal order Generalized Minimum Variance controller with a long-range prediction horizon effectively reduced controller and plant output variances.
• An increased prediction horizon demonstrated a significant ability to mitigate stochastic disturbances and attenuate oscillations.
• The controller performed well on both the water flow rate and electronic circuit systems under an incremental control scheme.
• The proposed tuning procedures provided effective means for optimizing controller performance.

Conclusions:

• The proposed Generalized Minimum Variance controller with a long-range prediction horizon offers an effective solution for disturbance mitigation and oscillation attenuation.
• The controller design and evaluation demonstrate its practical applicability in various process control scenarios.
• The study highlights the benefits of incorporating a long-range prediction horizon in minimum variance control strategies.