Application of Support Vector Machine to Obtain the Dynamic Model of Proton-Exchange Membrane Fuel Cell

Area of Science:

• * Energy Systems Engineering
• * Electrical Engineering
• * Computational Modeling

Background:

• * Proton-exchange membrane fuel cells (PEMFCs) are crucial for large-scale energy storage due to their independence from geographical constraints.
• * Accurate modeling of PEMFCs is essential for understanding their dynamic behavior and optimizing operation.
• * Existing models may struggle with the nonlinearities and noise inherent in fuel cell systems.

Purpose of the Study:

• * To develop and validate a novel support vector machine (SVM) based model for PEMFCs.
• * To accurately estimate the static and dynamic voltage-current characteristics of PEMFCs across all operating regions.
• * To statistically evaluate the proposed SVM model against established Diffusive Global (DG) and Evolution Strategy (ES) models.

Main Methods:

• * Implementation of a Support Vector Machine (SVM) algorithm to model PEMFC behavior.
• * Capturing static and dynamic voltage-current (I-V) characteristics within three distinct operating regions.
• * Experimental validation using a Ballard Nexa® 1.2 kW fuel cell power module.
• * Comparative statistical analysis against Diffusive Global (DG) and Evolution Strategy (ES) models.

Main Results:

• * The proposed SVM model demonstrated high accuracy in estimating PEMFC voltage compared to real experimental data.
• * The model successfully captured both static and dynamic voltage-current characteristics.
• * Statistical evaluation confirmed the superiority of the SVM model over DG and ES-based models.

Conclusions:

• * The SVM-based model provides a highly accurate and effective method for estimating PEMFC performance.
• * This approach is robust in handling nonlinearities and noise, crucial for real-world fuel cell applications.
• * The developed model offers a valuable tool for PEMFC analysis, control, and optimization.