Dual RBFNNs-Based Model-Free Adaptive Control With Aspen HYSYS Simulation

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Nonlinear Dynamics

Background:

• Traditional control methods often rely on accurate system models, which are difficult to obtain for complex nonlinear systems.
• Model-free adaptive control (MFAC) offers an alternative by learning directly from operational data.
• Existing MFAC techniques may lack systematic methods for controller structure design.

Purpose of the Study:

• To propose a novel data-driven model-free adaptive control (MFAC) method utilizing dual radial basis function neural networks (RBFNNs).
• To establish a systematic controller design methodology based solely on input-output (I/O) data, eliminating the need for first-principle or identified plant models.
• To rigorously analyze and guarantee the stability of both the closed-loop control system and the RBFNN training process.

Main Methods:

• A new MFAC approach is developed for discrete-time nonlinear systems.
• Controller structure is determined via equivalent-dynamic-linearization of the ideal nonlinear controller.
• Controller parameters are tuned using pseudogradient information derived from plant I/O data.
• Dual Radial Basis Function Neural Networks (RBFNNs) are employed for system representation and control.

Main Results:

• The proposed method enables controller design directly from I/O data, applicable to unknown nonlinear systems.
• Rigorous theoretical analysis confirms the stability of the closed-loop system and the RBFNN training.
• Effectiveness and applicability were validated through a numerical example and Aspen HYSYS simulation of a distillation column.

Conclusions:

• The novel MFAC method with dual RBFNNs provides a robust and data-driven approach for controlling discrete-time nonlinear systems.
• The systematic controller design based on I/O data represents a significant advancement over model-based techniques.
• The method's practical utility is confirmed by successful simulation in a chemical process application.