RBFNDOB-based neural network inverse control for non-minimum phase MIMO system with disturbances.

Area of Science:

• Control Systems Engineering
• Artificial Intelligence in Engineering
• Process Control

Background:

• Multi-input-multi-output (MIMO) systems often exhibit non-minimum phase characteristics and are susceptible to internal and external disturbances.
• Traditional control strategies struggle with the inherent complexities and uncertainties in such systems, leading to performance degradation.

Purpose of the Study:

• To develop an adaptive control strategy for non-minimum phase MIMO systems with disturbances.
• To enhance system stability, decoupling, and disturbance rejection capabilities.
• To validate the proposed control strategy through a case study on a ball mill grinding circuit.

Main Methods:

• A neural network inverse controller (NNIC) was developed using a Radial Basis Function Network (RBFN) to identify the inverse model of a constructed pseudo-plant, accommodating parameter variations.
• The MIMO system was decoupled and linearized into independent Single-Input-Single-Output (SISO) systems by cascading the NNIC with the original plant.
• A Radial Basis Function Network-based disturbance observer (RBFNDOB) was employed to estimate external disturbances, providing feed-forward compensation.

Main Results:

• The proposed NNIC-RBFNDOB strategy successfully decoupled and linearized the non-minimum phase MIMO system.
• The RBFNDOB effectively observed and compensated for external disturbances, improving control accuracy.
• Simulation results demonstrated the superior performance of the proposed adaptive control strategy compared to existing methods.

Conclusions:

• The combined NNIC and RBFNDOB adaptive control strategy offers a robust solution for controlling complex MIMO systems with non-minimum phase dynamics and disturbances.
• The method's effectiveness is validated by its successful application to a ball mill grinding circuit, showing significant improvements in performance.
• This approach provides a valuable tool for enhancing the stability and efficiency of industrial process control systems.