Predicting porosity in composite high-pressure hydrogen vessels using augmented fuzzy cognitive AI and manufacturing...

Area of Science:

• Materials Science
• Mechanical Engineering
• Artificial Intelligence

Background:

• High-pressure hydrogen storage vessels (HPV) are critical for hydrogen energy but susceptible to internal defects like porosity.
• Porosity can compromise the structural integrity of carbon fiber-reinforced composite HPVs, posing safety risks.

Purpose of the Study:

• To predict porosity number and rate category in Type IV HPVs using the XTRACTIS (XTS) artificial intelligence system.
• To identify key manufacturing variables influencing porosity formation and develop interpretable models for quality control.

Main Methods:

• Application of XTRACTIS (XTS), a general reasoning AI, for predictive modeling of porosity in composite HPVs.
• Development of fuzzy IF…THEN rules by XTS, selecting 15 out of 58 manufacturing variables to model porosity number.
• Evaluation of model performance using RMSE and correlation on an external test dataset.

Main Results:

• XTS successfully modeled the log10 of porosity number with an RMSE of 7.94% and a correlation of 0.824.
• Interpretable rules revealed that uniform fiber tension, optimized mandrel speed, and controlled winding parameters reduce porosity.
• AI identified critical process-defect relationships, highlighting the impact of fiber tension, winding angle, and layer volume.

Conclusions:

• Augmented fuzzy cognitive AI (XTS) effectively uncovers interpretable, domain-relevant interactions in composite manufacturing.
• The findings support enhanced quality control and process optimization for reducing defects in HPVs.
• Understanding these interactions is crucial for ensuring the safety and reliability of hydrogen storage systems.