Hybrid model using bond graph-TCN network and event triggered predictive control of pH neutralization process

Area of Science:

• Chemical Engineering
• Control Systems
• Artificial Intelligence

Background:

• pH neutralization is critical in industrial process control.
• Real-time pH neutralization units exhibit high complexity and nonlinearity.
• Existing models struggle with the dynamic uncertainties inherent in these systems.

Purpose of the Study:

• To introduce a novel fuzzy-based hybrid Bond Graph-Temporal Convolution Network (BG-TCN) model.
• To address the challenges posed by the convoluted dynamics of pH neutralization.
• To improve the precision and robustness of industrial process control loops.

Main Methods:

• Utilizing Temporal Convolutional Networks (TCN) with dilated causal convolutions for time series analysis.
• Employing Bond Graphs (BG) for an energy-centric representation of system dynamics.
• Integrating a fuzzy rule-based inference system to manage model uncertainties and facilitate smooth transitions.
• Implementing an event-triggered predictive control strategy with a fuzzy event handler.

Main Results:

• The hybrid BG-TCN model demonstrated superior performance compared to individual TCN and BG models.
• The model effectively handles the high nonlinearity and complexity of pH neutralization processes.
• Precise set point tracking was achieved for both closed-loop servo and regulatory control problems.

Conclusions:

• The proposed fuzzy-based BG-TCN hybrid model offers a significant advancement in pH neutralization control.
• This approach provides a robust and accurate solution for complex industrial applications.
• The integration of BG, TCN, and fuzzy logic enhances predictive control capabilities.