AI Replaces Humans in Identifying Causes of Fuel C

Newswise — Dr. Chi-Younger Jung’s analysis group from the Hydrogen Analysis & Demonstration Heart on the Korea Institute of Vitality Analysis (KIER) has efficiently developed a technique to research the microstructure of carbon fiber paper, a key materials in hydrogen gas cells, at a pace 100 instances quicker than present strategies. This was achieved by using digital twin expertise and synthetic intelligence (AI) studying.

Carbon fiber paper is a key materials in hydrogen gas cell stacks, enjoying an important function in facilitating water discharge and gas provide. It’s composed of supplies resembling carbon fibers, binders (adhesives), and coatings. Over time, the association, construction, and coating situation of those supplies change, resulting in a decline within the efficiency of the gas cell. Because of this, analyzing the microstructure of carbon fiber paper has develop into a vital step in diagnosing the situation of gas cells.

Nevertheless, real-time evaluation of the high-resolution microstructure of carbon fiber paper has been unattainable till now. It is because acquiring correct evaluation outcomes requires a course of by which the carbon fiber paper pattern is broken after which subjected to detailed examination utilizing an electron microscope.

To deal with the constraints of present evaluation strategies, the analysis group developed a expertise that analyzes the microstructure of carbon fiber paper utilizing X-ray diagnostics and an AI-based picture studying mannequin. Notably, this expertise permits exact evaluation utilizing solely X-ray tomography, eliminating the necessity for an electron microscope. In consequence, it permits for close to real-time situation analysis.

The analysis group extracted 5,000 photographs from over 200 samples of carbon fiber paper and educated a machine studying algorithm with this information. In consequence, the educated mannequin was capable of predict the 3D distribution and association of the important thing elements of carbon fiber paper — together with carbon fibers, binders, and coatings — with an accuracy of over 98%. This functionality permits the comparability of the preliminary state of the carbon fiber paper with its present state, permitting for the speedy identification of efficiency degradation causes.

The standard evaluation methodology, which includes crushing carbon fiber paper samples and utilizing an electron microscope, takes no less than 2 hours to finish. In distinction, the evaluation mannequin developed by the analysis group can determine the degradation, broken areas, and extent of harm within the carbon fiber paper inside a number of seconds utilizing solely X-ray tomography gear.

As well as, the analysis group utilized information from the developed mannequin to systematically determine how design components such because the thickness of the carbon fiber paper and the binder content material have an effect on gas cell efficiency. Additionally they extracted optimum design parameters and proposed a really perfect design plan aimed toward bettering the effectivity of gas cells.

Dr. Chi-Younger Jung, the lead researcher, acknowledged, “This examine is important in that it enhances evaluation expertise by combining AI with digital house utilization, and clearly identifies the connection between the construction and properties of power supplies, thereby demonstrating its sensible applicability.” He added, “We count on it to play a major function in associated fields resembling secondary batteries and water electrolysis sooner or later.“

This examine was performed with the help of the Korea Institute of Energy Research’s (KIER) analysis program and was revealed on-line in October 2024 in Utilized Vitality, a globally famend journal within the power sector.

Sensi Tech Hub
Logo