An Intelligent Condition-Monitoring Framework for Alkaline Water Electrolyzers Based on Hybrid Physics-Informed...

Area of Science:

• Electrochemical Engineering
• Artificial Intelligence
• Renewable Energy Systems

Background:

• Alkaline Water Electrolyzers (AWEs) are crucial for green hydrogen production.
• Operational risks arise from volatile renewable energy sources impacting AWE stability.
• Monitoring internal states of AWEs is challenging for ensuring reliable operation.

Purpose of the Study:

• To develop an intelligent condition-monitoring framework for AWEs.
• To address the challenge of inaccessible internal states in AWEs.
• To enhance the safety and efficiency of green hydrogen production.

Main Methods:

• A hybrid physics-informed machine learning (ML) methodology was employed.
• A high-fidelity Computational Fluid Dynamics (CFD) model was developed and validated.
• The CFD model generated multiphysics simulation data for training ML models.

Main Results:

• Eight key operational parameters were derived as Health Indicators (HIs).
• Three ML models were trained and benchmarked for health state classification.
• The Multilayer Perceptron (MLP) model achieved 90.43% accuracy.

Conclusions:

• The proposed framework enables reliable HI construction for AWEs.
• AI-enhanced condition monitoring is viable for improving AWE safety and efficiency.
• This approach supports the advancement of green hydrogen production technologies.