Data-driven set-point learning control with ESO and RBFNN for nonlinear batch processes subject to nonrepetitive...

Area of Science:

• Chemical Engineering
• Control Systems
• Process Optimization

Background:

• Nonlinear batch processes often face challenges with nonrepetitive uncertainties.
• Traditional control methods struggle with unknown dynamics and unmodeled disturbances.
• Data-driven approaches offer a promising alternative for complex process control.

Purpose of the Study:

• To develop a data-driven set-point learning control (DDSPLC) scheme for nonlinear batch processes.
• To address nonrepetitive uncertainties using only available process input and output data.
• To achieve robust batch optimization through adaptive set-point regulation.

Main Methods:

• Utilizing an extended state observer (ESO) for estimating unknown dynamics and disturbances.
• Employing an iterative dynamic linearization data model (IDLDM) to represent process behavior.
• Implementing a radial basis function neural network for estimating process information.
• Designing an adaptive set-point learning control law for closed-loop system optimization.

Main Results:

• The proposed DDSPLC scheme effectively handles nonrepetitive uncertainties in nonlinear batch processes.
• Robust convergence of output tracking error along the batch direction is rigorously proven.
• The method demonstrates effectiveness and advantages over existing approaches through validation examples.

Conclusions:

• The developed ESO-based DDSPLC scheme provides an effective data-driven solution for nonlinear batch process control.
• This approach offers a practical way to optimize batch processes using readily available data.
• The study validates the robustness and performance of the proposed control strategy.