UNIST Develops Dream Metal Material for Future Mobility Using 'Explainable Artificial Intelligence'

A high-strength lightweight alloy that can be used for personal flying vehicles and more was designed using XAI technology.

[Asia Economy Yeongnam Reporting Headquarters, Reporter Hwang Dooyul] A technology has been developed that uses artificial intelligence to design the elemental composition and manufacturing processes of alloys used in personal aerial vehicles (PAVs) and ultra-high-speed trains.

A research team led by Professor Jeong Imdu of the Department of Mechanical Engineering at UNIST has developed a new high-strength, lightweight aluminum alloy design technology using AI.

The research was conducted jointly with Gyeongsang National University, Korea Institute of Materials Science, and POSTECH.

The alloy material created with the elemental combinations and processes discovered by AI exhibited over 20% higher strength and more than 2.5 times greater ductility compared to the existing commercial material (7068-T6 alloy).

The technology incorporates 'Explainable Artificial Intelligence (XAI)' techniques.

This allows understanding of why AI recommended specific combinations and processes, enabling application in the development of various future mobility alloy materials.

The general characteristic that material strength and ductility are inversely proportional has been a major obstacle in the numerous attempts to find the ideal material that is both stronger and easier to process.

When designing alloys, it is necessary to find the optimal mixing ratio of additive elements and process conditions that provide high strength while maintaining sufficient ductility, but experimentally determining these ratios and conditions consumes enormous time and cost.

The joint research team developed a deep learning AI model that quickly identifies the optimal additive element combinations and process conditions for strength and ductility.

Using a recommendation algorithm, they also obtained process conditions predicted to yield excellent mechanical properties in the alloy.

The recommendation process takes less than five minutes, allowing designers to quickly obtain desired process conditions without complex and time-consuming experiments.

Following the new chemical compositions and process conditions recommended by AI, actual 7000-series aluminum alloys were fabricated, resulting in high-strength alloys maintaining a yield strength above 710 MPa and ductility of 20%.

Widely used commercial materials typically have a yield strength of about 590 MPa and ductility around 8%.

The advantage of applying 'explainable artificial intelligence technology' is that alloy design engineers can quantitatively understand how chemical composition and process conditions affect the mechanical properties of the alloy.

Knowing why AI recommended specific combinations and processes increases the reliability of the AI model's results.

Analysis of the microstructure of alloys actually produced based on AI recommendations confirmed that the interpretations of the 'explainable algorithm' align well with actual materials engineering theories.

First author student Park Seobin stated, “This technology can be broadly applied not only to aluminum alloys but also to the production of other lightweight alloy materials, and is expected to drastically reduce material development time and costs.”

Co-corresponding author Professor Seong Hyokyung of Gyeongsang National University said, “The ability to understand the key factors that strengthen materials through explainable AI enhances the technology’s reliability and applicability.”

Professor Jeong Imdu, the corresponding author who led the research, said, “We discovered lightweight metals with high mechanical properties that were difficult to find through experimental methods alone by integrating explainable AI research. This will be a core technology that maximizes safety while meeting the increasing demand for vehicle body lightweighting.”

The research results were published in January in the ‘Journal of Alloys and Compounds,’ an international journal ranked in the top 7% of JCR in the metals field.

This research was supported by the National Research Foundation of Korea, the Ministry of Trade, Industry and Energy, and the Korea Evaluation Institute of Industrial Technology (KEIT).

This content was produced with the assistance of AI translation services.